Data-intensive PhDs at LIV.INNO prepare students for careers outside of academia

1
00:00:07,679 --> 00:00:10,559
Hello, and welcome to the Physics World Weekly

2
00:00:10,559 --> 00:00:11,059
podcast,

3
00:00:11,359 --> 00:00:13,139
which is sponsored by LiveInno,

4
00:00:13,945 --> 00:00:17,565
the Liverpool Centre For Doctoral Training For Innovation

5
00:00:17,945 --> 00:00:20,045
in Data Intensive Science.

6
00:00:20,744 --> 00:00:22,605
The UK based CDT

7
00:00:23,064 --> 00:00:23,964
is a partnership

8
00:00:24,344 --> 00:00:27,864
between the University of Liverpool and Liverpool John

9
00:00:27,864 --> 00:00:28,844
Moores University.

10
00:00:29,980 --> 00:00:30,960
Since 2022,

11
00:00:31,820 --> 00:00:32,320
LiveInno

12
00:00:32,780 --> 00:00:35,200
has offered 3 cohorts of students,

13
00:00:35,579 --> 00:00:38,079
4 year, fully funded PhD

14
00:00:38,460 --> 00:00:38,960
studentships

15
00:00:39,500 --> 00:00:41,520
in data intensive science.

16
00:00:42,059 --> 00:00:44,559
From medical physics to quantum computing,

17
00:00:44,875 --> 00:00:47,615
the CDT has a diverse portfolio

18
00:00:47,994 --> 00:00:49,375
of research projects,

19
00:00:49,994 --> 00:00:53,215
but students all share training in high performance

20
00:00:53,354 --> 00:00:53,854
computing,

21
00:00:54,314 --> 00:00:56,255
machine learning, and AI,

22
00:00:56,635 --> 00:00:58,015
and data analysis.

23
00:00:58,700 --> 00:00:59,359
The CDT

24
00:00:59,740 --> 00:01:03,100
also gives students access to career advice and

25
00:01:03,100 --> 00:01:04,959
training in project management,

26
00:01:05,500 --> 00:01:06,000
entrepreneurship,

27
00:01:06,459 --> 00:01:07,280
and communication

28
00:01:07,579 --> 00:01:08,079
skills.

29
00:01:08,620 --> 00:01:11,120
As well as working on their original research,

30
00:01:11,555 --> 00:01:14,055
each student undertakes a 6 month

31
00:01:14,355 --> 00:01:15,495
industrial placement,

32
00:01:15,875 --> 00:01:19,174
and some projects are sponsored by an industrial

33
00:01:19,234 --> 00:01:20,295
partner throughout.

34
00:01:20,915 --> 00:01:24,240
The next cohort of PhD projects with a

35
00:01:24,240 --> 00:01:25,700
start date of autumn

36
00:01:26,079 --> 00:01:26,579
2025

37
00:01:27,439 --> 00:01:28,740
will be announced soon.

38
00:01:29,120 --> 00:01:31,859
More details can be found at www.liveinno.org.

39
00:01:36,284 --> 00:01:39,724
This podcast is hosted by Physics World's Katherine

40
00:01:39,724 --> 00:01:42,625
Schipper. And here she is in conversation

41
00:01:43,004 --> 00:01:44,545
with 2 LiveInno

42
00:01:45,004 --> 00:01:45,504
directors.

43
00:01:52,760 --> 00:01:55,000
Today, I'm joined down the line by Carsten

44
00:01:55,000 --> 00:01:57,400
Welch and Andrea Font, who are the director

45
00:01:57,400 --> 00:01:59,980
and the deputy director, respectively, of LiveInno.

46
00:02:00,840 --> 00:02:03,079
And as we as we record this, it's

47
00:02:03,079 --> 00:02:05,400
almost exactly a year since I submitted my

48
00:02:05,400 --> 00:02:08,014
PhD thesis. And as you can tell, I

49
00:02:08,014 --> 00:02:10,034
did not do a postdoc after my PhD.

50
00:02:10,574 --> 00:02:12,335
And actually that's true of most of my

51
00:02:12,335 --> 00:02:14,194
friends, who were in my cohort.

52
00:02:15,775 --> 00:02:16,275
So

53
00:02:16,574 --> 00:02:19,534
university research runs on the work of a

54
00:02:19,534 --> 00:02:21,615
lot of PhD students. And today we're going

55
00:02:21,615 --> 00:02:22,754
to be talking about

56
00:02:23,449 --> 00:02:25,710
what the value of PhD is for students

57
00:02:25,849 --> 00:02:27,849
when many of them will be at least

58
00:02:27,849 --> 00:02:30,189
be considering working outside of academia.

59
00:02:31,129 --> 00:02:32,650
When I think of most of the people

60
00:02:32,650 --> 00:02:34,729
I know who did PhDs, a lot of

61
00:02:34,729 --> 00:02:36,030
them are now data scientists,

62
00:02:36,834 --> 00:02:39,435
you know, which means that they're using data

63
00:02:39,435 --> 00:02:42,155
intensive analysis tools. They're using machine learning. They

64
00:02:42,155 --> 00:02:43,694
might be working with AI.

65
00:02:43,995 --> 00:02:45,694
And actually, even those

66
00:02:46,155 --> 00:02:48,155
of my friends who I know who are

67
00:02:48,155 --> 00:02:48,894
doing postdocs,

68
00:02:49,569 --> 00:02:51,409
a lot of them are also doing this

69
00:02:51,409 --> 00:02:53,409
data intensive science. They're doing the kind of

70
00:02:53,409 --> 00:02:55,030
work that has been transformed

71
00:02:55,409 --> 00:02:57,189
by AI and machine learning.

72
00:02:58,370 --> 00:03:00,290
So if we're talking about the value of

73
00:03:00,290 --> 00:03:01,189
doing a PhD,

74
00:03:01,585 --> 00:03:04,324
I think it seems it's important to ask

75
00:03:04,784 --> 00:03:07,344
whether universities are doing enough to give PhD

76
00:03:07,344 --> 00:03:08,885
students those kinds of skills.

77
00:03:09,504 --> 00:03:11,824
I am speaking to Carsten and Andrea today

78
00:03:11,824 --> 00:03:14,064
about the topic because LiVE. InO is geared

79
00:03:14,064 --> 00:03:17,129
towards students who are doing data intensive research

80
00:03:17,129 --> 00:03:17,629
projects

81
00:03:18,409 --> 00:03:20,349
and who are at least considering

82
00:03:20,730 --> 00:03:24,010
using their scientific skills outside of academia in

83
00:03:24,010 --> 00:03:24,510
industry.

84
00:03:25,290 --> 00:03:27,914
So Carsten Walsh is based at the University

85
00:03:27,914 --> 00:03:30,155
of Liverpool, and his research is on the

86
00:03:30,155 --> 00:03:32,735
design of particle accelerators and light sources.

87
00:03:33,194 --> 00:03:36,314
Andrea Font is a computational astrophysicist at Liverpool

88
00:03:36,314 --> 00:03:37,294
John Moores University

89
00:03:37,675 --> 00:03:39,675
who studies the formation of the Milky Way

90
00:03:39,675 --> 00:03:40,175
galaxy.

91
00:03:40,794 --> 00:03:42,655
Thank you so much for joining me today.

92
00:03:43,269 --> 00:03:44,729
Yeah. It's great to be here.

93
00:03:45,269 --> 00:03:47,930
So, Carsten, I'm gonna start with a nice

94
00:03:48,150 --> 00:03:50,629
easy question. What is the purpose of doing

95
00:03:50,629 --> 00:03:51,209
a PhD?

96
00:03:52,549 --> 00:03:53,049
Well,

97
00:03:53,430 --> 00:03:55,189
I'm not sure if that's such an easy

98
00:03:55,189 --> 00:03:57,754
question. I think the the purpose of the

99
00:03:57,754 --> 00:04:00,634
PhD, the primary purpose really is the to

100
00:04:00,634 --> 00:04:01,935
demonstrate the ability

101
00:04:02,474 --> 00:04:03,854
to conduct independent

102
00:04:04,155 --> 00:04:04,655
research.

103
00:04:05,034 --> 00:04:07,594
Now that sounds simple, but it's really not

104
00:04:07,594 --> 00:04:10,094
a small step up from the 1st degree.

105
00:04:10,340 --> 00:04:12,900
It's in fact a step that requires very

106
00:04:12,900 --> 00:04:13,400
significant

107
00:04:14,019 --> 00:04:17,139
skills development. And these are skills not just

108
00:04:17,139 --> 00:04:20,180
in terms of additional research skills, but very

109
00:04:20,180 --> 00:04:21,720
importantly, also skills

110
00:04:22,100 --> 00:04:23,800
in wider professional skills.

111
00:04:24,235 --> 00:04:25,375
Now also,

112
00:04:25,914 --> 00:04:26,975
having said this,

113
00:04:27,354 --> 00:04:29,694
I guess the ideal PhD training

114
00:04:30,235 --> 00:04:30,735
should

115
00:04:31,194 --> 00:04:33,754
consider more than just one possible career pathway

116
00:04:33,754 --> 00:04:35,675
as you said at the beginning. So the

117
00:04:35,675 --> 00:04:38,069
question that is also, how do you train

118
00:04:38,069 --> 00:04:40,629
students for the career that they would like

119
00:04:40,629 --> 00:04:41,209
to do?

120
00:04:41,669 --> 00:04:43,990
That, I think, triggers immediately the question, what

121
00:04:43,990 --> 00:04:45,909
gives us the best research? So how can

122
00:04:45,909 --> 00:04:46,409
we,

123
00:04:46,709 --> 00:04:48,709
get the best research out of what we

124
00:04:48,709 --> 00:04:51,604
are doing? And to me, there are 3

125
00:04:51,604 --> 00:04:54,404
key ingredients that are needed. 1st, it needs

126
00:04:54,404 --> 00:04:56,745
an academic environment, an environment

127
00:04:57,365 --> 00:05:00,644
where challenging questions can be asked freely and

128
00:05:00,644 --> 00:05:03,544
where people are not afraid of thinking thinking

129
00:05:03,604 --> 00:05:06,800
creatively well beyond the current status quo.

130
00:05:07,180 --> 00:05:08,959
It also need access to

131
00:05:09,419 --> 00:05:13,100
international and national research laboratories, so large scale

132
00:05:13,100 --> 00:05:16,139
research infrastructure where people can do the big

133
00:05:16,139 --> 00:05:17,199
science experiments.

134
00:05:17,579 --> 00:05:20,524
And, certainly, it needs industry to basically pull

135
00:05:20,524 --> 00:05:23,404
the academics back to the ground and to

136
00:05:23,404 --> 00:05:23,904
reality,

137
00:05:24,685 --> 00:05:25,345
to really

138
00:05:25,725 --> 00:05:28,365
drive research and innovation in a way that

139
00:05:28,365 --> 00:05:29,664
can benefit society.

140
00:05:30,204 --> 00:05:31,504
And all these ingredients,

141
00:05:32,229 --> 00:05:34,389
they have formed the training that we provide

142
00:05:34,389 --> 00:05:35,449
in Live in Hope.

143
00:05:36,389 --> 00:05:38,229
Right. And so at the end there, you

144
00:05:38,229 --> 00:05:39,370
mentioned bringing

145
00:05:39,909 --> 00:05:40,409
academics

146
00:05:40,709 --> 00:05:42,310
down to the ground. But Andrea, as an

147
00:05:42,310 --> 00:05:43,930
astrophysicist, you're obviously

148
00:05:44,504 --> 00:05:45,004
looking,

149
00:05:45,625 --> 00:05:48,025
you know, far far into the far into

150
00:05:48,025 --> 00:05:50,444
the galaxy. And in in astrophysics,

151
00:05:50,745 --> 00:05:53,324
you generate a a vast amount of data

152
00:05:53,465 --> 00:05:56,264
from observations and simulations that you have to

153
00:05:56,264 --> 00:05:56,764
analyze.

154
00:05:57,379 --> 00:05:59,620
And I'm expecting that the amount of data

155
00:05:59,620 --> 00:06:00,279
you generate

156
00:06:00,659 --> 00:06:01,639
is only increasing.

157
00:06:02,019 --> 00:06:04,360
So can you talk a bit about what

158
00:06:04,579 --> 00:06:06,680
that looks like in your field and how

159
00:06:07,459 --> 00:06:08,360
data intensive

160
00:06:09,139 --> 00:06:11,939
tools like machine learning and AI are being

161
00:06:11,939 --> 00:06:13,185
used in your field?

162
00:06:13,665 --> 00:06:16,305
So, yeah, as as you said, astronomy needs

163
00:06:16,305 --> 00:06:17,605
huge amounts of data.

164
00:06:18,064 --> 00:06:19,985
This is a vast universe, so we need

165
00:06:19,985 --> 00:06:21,285
lots and lots of data.

166
00:06:22,464 --> 00:06:23,105
There is,

167
00:06:23,504 --> 00:06:25,584
to give an example of a of a

168
00:06:25,584 --> 00:06:26,084
current

169
00:06:26,449 --> 00:06:29,509
survey, Euclid, which is recently launched

170
00:06:30,050 --> 00:06:30,629
by ESA.

171
00:06:31,490 --> 00:06:32,550
Euclid is

172
00:06:32,930 --> 00:06:33,430
surveying

173
00:06:33,810 --> 00:06:35,350
2 thirds of the sky,

174
00:06:35,889 --> 00:06:38,550
of the entire sky, and it sends back

175
00:06:39,584 --> 00:06:43,045
approximately 100 gigabytes of data every single night

176
00:06:43,185 --> 00:06:44,785
and it will do so for the next

177
00:06:44,785 --> 00:06:47,185
6 years or so. So this is just

178
00:06:47,185 --> 00:06:50,464
one example of survey but there's many other

179
00:06:50,464 --> 00:06:52,920
surveys out there. They're surveying the sky in

180
00:06:52,920 --> 00:06:55,180
multiple wavelengths, so we have a multidimensional

181
00:06:55,720 --> 00:06:56,459
data set.

182
00:06:57,079 --> 00:06:58,459
So in this vast

183
00:06:58,920 --> 00:07:01,879
data set, we have millions and millions of

184
00:07:01,879 --> 00:07:02,379
galaxies,

185
00:07:03,160 --> 00:07:05,879
objects that are moving very fast across the

186
00:07:05,879 --> 00:07:06,379
sky,

187
00:07:06,714 --> 00:07:09,294
like asteroids and comets in the solar system,

188
00:07:09,754 --> 00:07:12,654
objects that vary in brightness, like supernovae

189
00:07:13,035 --> 00:07:14,574
or stars that pulsate.

190
00:07:15,115 --> 00:07:17,375
All of these objects need to be classified

191
00:07:18,154 --> 00:07:18,975
and categorized.

192
00:07:19,354 --> 00:07:21,754
So this is where machine learning is helping

193
00:07:21,754 --> 00:07:22,789
us because,

194
00:07:23,329 --> 00:07:25,649
with that huge amounts of data, it's it's

195
00:07:25,649 --> 00:07:28,389
impossible for humans to do all the work.

196
00:07:28,849 --> 00:07:30,470
So this is kind of the primary

197
00:07:30,930 --> 00:07:33,589
example of applying machine learning in astronomy.

198
00:07:34,529 --> 00:07:36,629
Okay. Thank you. It sounds like it's really

199
00:07:38,235 --> 00:07:40,235
accelerating the speed at which you can do

200
00:07:40,235 --> 00:07:43,035
these kinds of very data intensive analysis. Okay.

201
00:07:43,035 --> 00:07:45,535
And so, Andrea, you spoke about,

202
00:07:46,795 --> 00:07:48,714
some of the techniques that you're using in

203
00:07:48,714 --> 00:07:49,535
your field.

204
00:07:50,394 --> 00:07:50,894
And

205
00:07:52,029 --> 00:07:54,449
what I'm interested to know about is whether

206
00:07:54,750 --> 00:07:57,569
you think the physics community as a whole

207
00:07:57,790 --> 00:07:58,290
is

208
00:07:58,990 --> 00:08:01,490
adapting to the demands of

209
00:08:01,790 --> 00:08:04,394
data intensive research and the new skills that

210
00:08:04,794 --> 00:08:07,435
incoming PhD students are going to need to

211
00:08:07,435 --> 00:08:08,654
do this kind of work.

212
00:08:09,274 --> 00:08:10,175
And also

213
00:08:10,794 --> 00:08:12,894
whether the physics community is adapting

214
00:08:13,595 --> 00:08:14,894
enough to the fact that

215
00:08:15,274 --> 00:08:16,735
most PhD students,

216
00:08:17,189 --> 00:08:19,770
you know, who are doing this quite fundamental

217
00:08:19,990 --> 00:08:21,210
research will nevertheless

218
00:08:22,069 --> 00:08:24,330
go on to have careers outside of academia.

219
00:08:25,830 --> 00:08:28,470
Yeah, that's a good point. And, I think

220
00:08:28,470 --> 00:08:31,290
the physics and astrophysics communities are adapting

221
00:08:32,345 --> 00:08:34,904
quite well to the new challenges. Of course,

222
00:08:34,904 --> 00:08:35,965
there's a rapid

223
00:08:36,424 --> 00:08:38,285
pace of development in the field.

224
00:08:38,825 --> 00:08:40,125
I can speak for astronomy.

225
00:08:41,065 --> 00:08:43,799
It has been estimated that the number of

226
00:08:43,959 --> 00:08:47,080
publications that mentioned the word AI or machine

227
00:08:47,080 --> 00:08:49,799
learning or proposed new methods in these fields

228
00:08:49,799 --> 00:08:50,299
are

229
00:08:50,759 --> 00:08:53,659
doubling every year and a half in astronomy

230
00:08:53,799 --> 00:08:56,600
alone. So, as researchers, we need to keep

231
00:08:56,600 --> 00:08:57,980
up with all this new developments.

232
00:08:58,475 --> 00:09:01,195
Now how to incorporate that into training is

233
00:09:01,195 --> 00:09:03,855
kind of tricky, but it's doable.

234
00:09:04,634 --> 00:09:06,014
So in addition to

235
00:09:06,394 --> 00:09:09,035
more traditional training that we offer in terms

236
00:09:09,035 --> 00:09:10,095
of, like, courses

237
00:09:10,715 --> 00:09:11,855
in computer science

238
00:09:12,475 --> 00:09:14,399
and and, physics and astro.

239
00:09:15,440 --> 00:09:16,179
We also,

240
00:09:17,040 --> 00:09:19,220
adapt to the this fast

241
00:09:20,000 --> 00:09:20,500
pace

242
00:09:20,960 --> 00:09:23,120
by doing in house training with the new

243
00:09:23,120 --> 00:09:24,740
methods that are they're proposed.

244
00:09:26,879 --> 00:09:27,129
So,

245
00:09:28,205 --> 00:09:29,024
we also,

246
00:09:29,725 --> 00:09:31,404
so we could do that in variety of

247
00:09:31,404 --> 00:09:34,845
ways through journal clubs, summer schools, or training

248
00:09:34,845 --> 00:09:38,205
in in in partnership with our industry partners

249
00:09:38,205 --> 00:09:38,865
as well.

250
00:09:40,620 --> 00:09:43,120
So in addition to this, we are focusing

251
00:09:43,179 --> 00:09:43,679
on,

252
00:09:44,700 --> 00:09:46,940
giving the new skills and the new economy.

253
00:09:46,940 --> 00:09:49,580
We're also focusing on, so what we call,

254
00:09:49,580 --> 00:09:50,879
transformative skills.

255
00:09:51,740 --> 00:09:53,679
We always been aware that the students

256
00:09:55,154 --> 00:09:58,134
build these transformative skills in during their research.

257
00:09:58,514 --> 00:10:00,215
What I'm talking here is about

258
00:10:00,675 --> 00:10:02,455
communication skills, networking,

259
00:10:04,835 --> 00:10:08,355
and more recently developing more entrepreneurial skills as

260
00:10:08,355 --> 00:10:11,409
well in addition to the more standard data

261
00:10:11,470 --> 00:10:11,970
skills.

262
00:10:13,069 --> 00:10:13,569
So,

263
00:10:14,029 --> 00:10:16,350
but now it's becoming more focused because the

264
00:10:16,350 --> 00:10:19,309
students need to interact with industry partners so

265
00:10:19,309 --> 00:10:21,649
they they actually can see for themselves how

266
00:10:22,095 --> 00:10:24,574
the value of these transformative skills in the

267
00:10:24,574 --> 00:10:26,495
real world. Okay, thank you. Carsten, do you

268
00:10:26,495 --> 00:10:27,875
have anything to add to that?

269
00:10:29,134 --> 00:10:31,154
Well, I think when it comes to how

270
00:10:31,454 --> 00:10:34,334
skills have developed, the key skills that students

271
00:10:34,334 --> 00:10:36,419
need, it's fair to say that this has

272
00:10:36,419 --> 00:10:39,700
been an incredibly interesting journey over maybe the

273
00:10:39,700 --> 00:10:40,919
past 10 years.

274
00:10:41,620 --> 00:10:44,120
Personally, I had the pleasure of coordinating

275
00:10:44,580 --> 00:10:48,179
6 Marie Career Networks across life sciences, physics,

276
00:10:48,179 --> 00:10:48,919
and engineering,

277
00:10:49,595 --> 00:10:51,754
as well as the 2 doctoral training centers

278
00:10:51,754 --> 00:10:53,995
here. And I was also the chair of

279
00:10:53,995 --> 00:10:54,495
SDFC's

280
00:10:54,875 --> 00:10:58,095
education, training, and careers committee for several years.

281
00:10:58,315 --> 00:11:00,154
And in combination, all of this has allowed

282
00:11:00,154 --> 00:11:01,695
me to really think quite carefully

283
00:11:02,154 --> 00:11:03,149
about what makes

284
00:11:03,629 --> 00:11:06,450
the ideal training for an early stage researcher,

285
00:11:06,509 --> 00:11:07,250
in particular

286
00:11:07,550 --> 00:11:11,009
for PhD students, and what role do cohort

287
00:11:11,070 --> 00:11:12,050
based approaches

288
00:11:12,590 --> 00:11:15,230
play in this as compared to individual single

289
00:11:15,230 --> 00:11:17,570
PhDs done at any university.

290
00:11:18,095 --> 00:11:19,535
And I think it's it's fair to say

291
00:11:19,535 --> 00:11:21,054
that today we live in one of the

292
00:11:21,054 --> 00:11:23,295
most dynamic times when it comes to the

293
00:11:23,295 --> 00:11:26,495
skills that are required today to do well

294
00:11:26,495 --> 00:11:27,235
in research.

295
00:11:27,695 --> 00:11:29,955
From changes to digital literacy,

296
00:11:30,575 --> 00:11:32,355
to how we work together,

297
00:11:32,769 --> 00:11:34,870
and also how we look after each other,

298
00:11:35,409 --> 00:11:37,649
all of these aspects, I think, have changed

299
00:11:37,649 --> 00:11:39,889
hugely over the past 5 years, and they

300
00:11:39,889 --> 00:11:43,350
keep evolving at a pace where national frameworks

301
00:11:43,809 --> 00:11:46,370
simply cannot follow quickly enough. So it's really

302
00:11:46,370 --> 00:11:49,705
down to training initiatives like CDTs,

303
00:11:50,644 --> 00:11:52,825
to pave the way for the next generation

304
00:11:52,884 --> 00:11:56,325
of researchers by bringing those skills into the

305
00:11:56,325 --> 00:11:56,825
training

306
00:11:57,205 --> 00:11:57,945
very dynamically

307
00:11:58,644 --> 00:12:00,565
when they are needed and the moment they

308
00:12:00,565 --> 00:12:02,440
come up. Now one of the aspects that

309
00:12:02,440 --> 00:12:04,679
has been very important in our case has

310
00:12:04,679 --> 00:12:06,379
been to exploit synergies.

311
00:12:07,000 --> 00:12:09,240
In Livinum, we exploit synergies all of the

312
00:12:09,240 --> 00:12:12,600
time between the different SDFC research communities. We

313
00:12:12,600 --> 00:12:14,379
have seen that there is benefit

314
00:12:14,725 --> 00:12:18,105
in bringing nuclear and particle physicists together with

315
00:12:18,404 --> 00:12:18,904
astrophysicists

316
00:12:19,365 --> 00:12:20,105
and accelerator

317
00:12:20,485 --> 00:12:23,684
experts, and that by doing this, everybody benefits

318
00:12:23,684 --> 00:12:25,845
and that the common denominator between all of

319
00:12:25,845 --> 00:12:28,470
these areas is data science. So this has

320
00:12:28,470 --> 00:12:29,529
been the foundation

321
00:12:29,830 --> 00:12:31,750
of Liv Inno since the start where we

322
00:12:31,750 --> 00:12:32,970
have built those bridges.

323
00:12:33,350 --> 00:12:34,250
But we've also

324
00:12:34,950 --> 00:12:38,149
organized trainings together with other large scale training

325
00:12:38,149 --> 00:12:41,429
initiatives, in particular, the Eupraxia doctoral network, which

326
00:12:41,429 --> 00:12:45,295
looks after plasma accelerator research. And by bringing

327
00:12:45,295 --> 00:12:48,835
PhD students from seemingly different communities together,

328
00:12:49,375 --> 00:12:52,175
we have enriched everyone's training, and it has

329
00:12:52,175 --> 00:12:55,134
turned out to be beneficial for everyone involved.

330
00:12:55,134 --> 00:12:57,269
It this has been so successful that it

331
00:12:57,269 --> 00:12:58,009
was recognized

332
00:12:58,389 --> 00:13:01,350
by the European Commission as an international success

333
00:13:01,350 --> 00:13:02,970
story and demonstrated

334
00:13:03,350 --> 00:13:05,610
as a way forward in researcher training.

335
00:13:06,149 --> 00:13:09,029
Now what have our students done, as part

336
00:13:09,029 --> 00:13:09,929
of their projects?

337
00:13:10,434 --> 00:13:13,235
In Livinor, every single PhD student has to

338
00:13:13,235 --> 00:13:15,315
do an industry placement, so they have to

339
00:13:15,315 --> 00:13:18,534
spend at least 6 months in industry working

340
00:13:18,754 --> 00:13:20,934
on a topic outside of their PhD.

341
00:13:21,394 --> 00:13:23,850
The aim is that they apply the skills

342
00:13:23,850 --> 00:13:26,089
that they have learned during their PhD onto

343
00:13:26,089 --> 00:13:27,470
a real world problem

344
00:13:27,929 --> 00:13:30,909
working with another sector. That has many benefits.

345
00:13:30,970 --> 00:13:32,889
It helps the other sector to solve their

346
00:13:32,889 --> 00:13:35,769
problems. It really enriches the students' training, and

347
00:13:35,769 --> 00:13:37,725
in particular, it also gives them a real

348
00:13:37,725 --> 00:13:40,125
idea of how their career might look like

349
00:13:40,125 --> 00:13:41,804
in the future. As you said at the

350
00:13:41,804 --> 00:13:42,945
beginning, many students

351
00:13:43,325 --> 00:13:45,424
do not end up in the academic sector.

352
00:13:45,725 --> 00:13:47,725
At the same time, most students do not

353
00:13:47,725 --> 00:13:49,404
really have an idea of how it is

354
00:13:49,404 --> 00:13:51,610
to work in industry. So we built this

355
00:13:51,610 --> 00:13:53,449
into our training, and we made sure that

356
00:13:53,449 --> 00:13:56,089
the students get exposure to the industry sector

357
00:13:56,089 --> 00:13:58,029
so that they can take an informed decision.

358
00:13:58,089 --> 00:13:59,449
And the things that they have done in

359
00:13:59,449 --> 00:14:00,110
these placements

360
00:14:00,490 --> 00:14:02,409
are amazing. To just give one example, we

361
00:14:02,409 --> 00:14:04,644
had a student who worked with a relatively

362
00:14:04,644 --> 00:14:05,464
small company

363
00:14:05,845 --> 00:14:08,745
on the data related challenges of a well-being

364
00:14:08,964 --> 00:14:11,625
being platform that was established during the pandemic.

365
00:14:11,845 --> 00:14:14,644
There was a platform targeting teenagers in particular

366
00:14:14,644 --> 00:14:17,444
that had mental health issues, and our student

367
00:14:17,444 --> 00:14:18,949
used her data skills,

368
00:14:19,649 --> 00:14:22,529
to develop a chat platform to analyze the

369
00:14:22,529 --> 00:14:24,949
data coming from the participants in the program

370
00:14:25,250 --> 00:14:27,250
and to really inform the way forward for

371
00:14:27,250 --> 00:14:28,389
that support platform.

372
00:14:28,690 --> 00:14:30,709
The results were amazing for everybody,

373
00:14:31,134 --> 00:14:34,574
the teenagers, our student who really loved working

374
00:14:34,574 --> 00:14:36,815
in that context, and also for the company

375
00:14:36,815 --> 00:14:39,214
who got somebody very skilled and talented to

376
00:14:39,214 --> 00:14:41,235
work on a data challenge that they had.

377
00:14:42,095 --> 00:14:43,554
Okay. Thanks. And so

378
00:14:44,149 --> 00:14:45,750
you you've spoken about, you know, one of

379
00:14:45,750 --> 00:14:48,389
these industry industry placements, and I'm sort of

380
00:14:48,389 --> 00:14:49,690
interested in the fact that

381
00:14:50,309 --> 00:14:51,750
as well as as well as, you know,

382
00:14:51,750 --> 00:14:53,830
they work with industry, but actually quite a

383
00:14:53,830 --> 00:14:56,389
lot of the students do what I what

384
00:14:56,389 --> 00:14:58,649
I would think of as as quite fundamental

385
00:14:58,950 --> 00:15:00,544
research or quite, you

386
00:15:01,644 --> 00:15:04,845
know, abstract maybe astrophysical or cosmological research. So

387
00:15:04,845 --> 00:15:06,445
actually, I have a question for Andrea, which

388
00:15:06,445 --> 00:15:08,065
is you spoke a bit about,

389
00:15:08,524 --> 00:15:10,704
at the start, about these techniques

390
00:15:11,164 --> 00:15:14,784
in machine learning that you use in astrophysics.

391
00:15:15,910 --> 00:15:17,610
And a lot of that was about,

392
00:15:19,110 --> 00:15:21,690
about learning patterns in simulations or about analyzing

393
00:15:22,230 --> 00:15:23,690
images from a telescope.

394
00:15:24,230 --> 00:15:25,290
And so I'm interested

395
00:15:25,750 --> 00:15:27,509
to maybe you have an example of how

396
00:15:27,509 --> 00:15:29,350
a student who is doing that kind of

397
00:15:29,350 --> 00:15:33,488
research that, you know, really, you know, really

398
00:15:33,528 --> 00:15:37,813
pure astrophysics, maybe cosmological research might end up

399
00:15:37,853 --> 00:15:42,138
using those skills in a context that is

400
00:15:42,178 --> 00:15:43,219
in industry.

401
00:15:44,299 --> 00:15:46,379
We have, we have a student that is,

402
00:15:46,700 --> 00:15:48,240
currently using cosmological

403
00:15:48,540 --> 00:15:49,040
simulations

404
00:15:49,660 --> 00:15:52,299
to understand the nature of dark matter. Now

405
00:15:52,299 --> 00:15:55,200
dark matter is not seen, but is everywhere.

406
00:15:55,259 --> 00:15:57,345
So it's not detectable, but it's everywhere. What

407
00:15:57,345 --> 00:16:00,464
is actually seen is luminous matter, which is

408
00:16:00,464 --> 00:16:02,725
actually quite sparse. So it's not

409
00:16:03,105 --> 00:16:05,365
the dark matter structures are not continuously

410
00:16:05,985 --> 00:16:09,504
or uniformly populated with luminous matter. So it's,

411
00:16:09,664 --> 00:16:10,884
in terms of statistics,

412
00:16:11,330 --> 00:16:13,029
it's a sparsely populated

413
00:16:13,570 --> 00:16:14,769
problem. So you have,

414
00:16:15,090 --> 00:16:17,169
you're trying to get to the nature of

415
00:16:17,169 --> 00:16:19,590
that matter, but you do that through the

416
00:16:19,649 --> 00:16:20,149
analyzing,

417
00:16:21,250 --> 00:16:22,309
sparsely populated,

418
00:16:23,250 --> 00:16:26,309
data points, which are basically luminous galaxies.

419
00:16:28,355 --> 00:16:30,454
So student developed techniques to,

420
00:16:31,314 --> 00:16:33,954
to do research in cosmology, but turns out

421
00:16:33,954 --> 00:16:36,534
that the similar techniques are actually have application

422
00:16:36,595 --> 00:16:38,674
in the real world and and and, for

423
00:16:38,674 --> 00:16:41,014
example, in in the Earth observation.

424
00:16:42,420 --> 00:16:45,540
Now with ops with data from satellites, there's

425
00:16:45,540 --> 00:16:47,860
many satellites out there that observe the Earth,

426
00:16:47,860 --> 00:16:50,420
but sometimes the data is incomplete. So, again,

427
00:16:50,420 --> 00:16:53,220
it's sparsely populated because of the income it

428
00:16:53,220 --> 00:16:55,620
would be for the cloud coverage or any

429
00:16:55,620 --> 00:16:56,759
other issues with

430
00:16:57,365 --> 00:17:00,325
incompleteness of the data. So he was able

431
00:17:00,325 --> 00:17:04,505
to use similar statistical methods to apply to,

432
00:17:05,204 --> 00:17:06,345
the Earth observation

433
00:17:06,724 --> 00:17:07,224
data

434
00:17:07,605 --> 00:17:09,384
to, for example, to improve

435
00:17:11,519 --> 00:17:12,340
the the landscaping,

436
00:17:12,720 --> 00:17:13,380
the the

437
00:17:13,840 --> 00:17:16,180
observations of the of the land to,

438
00:17:16,720 --> 00:17:18,740
with the real applications into agriculture

439
00:17:19,279 --> 00:17:22,559
and increasing crop efficiency, for example. So that

440
00:17:22,559 --> 00:17:25,539
was a very interesting result. So you unexpected

441
00:17:26,325 --> 00:17:27,544
application of cosmology

442
00:17:27,845 --> 00:17:28,744
to agriculture.

443
00:17:29,284 --> 00:17:30,744
So I know you're a you're a conversational

444
00:17:30,804 --> 00:17:33,284
astrophysicist, but I'm I'm also guessing that if

445
00:17:33,284 --> 00:17:33,944
you are

446
00:17:35,044 --> 00:17:37,544
an astrophysicist, and this is maybe true in

447
00:17:37,605 --> 00:17:39,924
in certain areas of accelerated physics as well,

448
00:17:39,924 --> 00:17:40,164
that

449
00:17:40,789 --> 00:17:42,569
and you get very good at, say, analyzing

450
00:17:42,630 --> 00:17:44,789
images with AI and machine learning. That's an

451
00:17:44,789 --> 00:17:46,950
incredibly in demand skill as well, isn't it,

452
00:17:46,950 --> 00:17:49,509
at the moment? Yes. Absolutely. So,

453
00:17:49,909 --> 00:17:51,529
as as Carson has mentioned,

454
00:17:51,990 --> 00:17:54,630
there's vast applications in health care, for example,

455
00:17:54,630 --> 00:17:55,609
where there's also

456
00:17:55,974 --> 00:17:58,855
lots of image processing there. So we actually

457
00:17:58,855 --> 00:18:00,875
have similar techniques in physics,

458
00:18:01,414 --> 00:18:01,914
astrophysics,

459
00:18:02,694 --> 00:18:03,754
and in other

460
00:18:04,054 --> 00:18:07,575
fields including healthcare. So there's similar applications. So

461
00:18:07,575 --> 00:18:09,914
this is again another point that students actually

462
00:18:10,934 --> 00:18:11,434
learn

463
00:18:13,690 --> 00:18:16,250
this transformative skills that I mentioned earlier. They

464
00:18:16,250 --> 00:18:18,730
actually have the skills to apply to the

465
00:18:18,730 --> 00:18:20,190
real world data,

466
00:18:20,890 --> 00:18:23,309
but actually having the placement and doing

467
00:18:24,970 --> 00:18:27,865
a separate project and applying these skills into

468
00:18:27,865 --> 00:18:30,345
an entirely different field, it just opens their

469
00:18:30,345 --> 00:18:32,904
eyes and opens different opportunities for them as

470
00:18:32,904 --> 00:18:35,144
well, also career wise as well. Yeah. And

471
00:18:35,144 --> 00:18:35,964
I have a

472
00:18:36,825 --> 00:18:39,404
question for for Carsten. And this is about

473
00:18:39,865 --> 00:18:41,544
I guess I guess going back again to

474
00:18:41,544 --> 00:18:42,204
the philosophy

475
00:18:42,585 --> 00:18:42,990
of

476
00:18:44,429 --> 00:18:45,089
the CDT.

477
00:18:45,710 --> 00:18:46,769
So I think

478
00:18:47,309 --> 00:18:47,809
if

479
00:18:48,269 --> 00:18:51,549
you're working with, say, a single supervisor, your

480
00:18:51,549 --> 00:18:53,309
experience of the PhD and the way you

481
00:18:53,309 --> 00:18:54,849
see yourself in your project

482
00:18:55,565 --> 00:18:56,065
depends

483
00:18:56,605 --> 00:18:58,605
very much on the supervisor you get. And

484
00:18:58,605 --> 00:18:59,984
by that I mean whether

485
00:19:00,444 --> 00:19:01,184
you are,

486
00:19:02,444 --> 00:19:04,125
whether they are your boss or whether they're

487
00:19:04,125 --> 00:19:05,724
sort of more like your collaborator. And I

488
00:19:05,724 --> 00:19:07,744
think there's a big spectrum there.

489
00:19:08,089 --> 00:19:11,309
And so how do you see the students

490
00:19:11,369 --> 00:19:13,369
that come in to the CDT? What do

491
00:19:13,369 --> 00:19:14,910
you think is the best way for

492
00:19:15,609 --> 00:19:18,890
supervisors or people who are organizing CDTs to

493
00:19:18,890 --> 00:19:19,390
view

494
00:19:20,535 --> 00:19:22,555
their students? And how do

495
00:19:23,015 --> 00:19:24,455
you how have you thought about that when

496
00:19:24,455 --> 00:19:26,475
you've been designing these training,

497
00:19:26,934 --> 00:19:28,855
this training that you give the students on

498
00:19:28,855 --> 00:19:29,914
that live in OCD?

499
00:19:31,015 --> 00:19:34,295
Yeah. So there's definitely not one way that

500
00:19:34,295 --> 00:19:37,690
works for every student. Every arrangement is always

501
00:19:37,690 --> 00:19:38,190
bespoke

502
00:19:38,970 --> 00:19:41,529
and what we do is we establish a

503
00:19:41,529 --> 00:19:44,490
structured career development plan at the beginning of

504
00:19:44,490 --> 00:19:45,309
every PhD

505
00:19:45,930 --> 00:19:48,695
project and that is a document which covers

506
00:19:48,695 --> 00:19:51,275
the research, but also the training, the anticipated

507
00:19:51,894 --> 00:19:55,174
workshop participation, conferences, publications, all of the things

508
00:19:55,174 --> 00:19:57,174
that make a PhD. So it's quite a

509
00:19:57,174 --> 00:19:58,154
complex document.

510
00:19:58,535 --> 00:20:00,535
And in that document, we have a dialogue

511
00:20:00,535 --> 00:20:02,394
between the student and the supervisory

512
00:20:02,695 --> 00:20:05,420
team, and they write up at the start

513
00:20:05,420 --> 00:20:06,400
what they think

514
00:20:06,779 --> 00:20:09,180
is the most ambitious but also the most

515
00:20:09,180 --> 00:20:10,320
enjoyable combination,

516
00:20:11,900 --> 00:20:14,480
for the the next 3 or 4 years.

517
00:20:14,539 --> 00:20:16,539
And then we take that document, and we

518
00:20:16,539 --> 00:20:18,494
update it every 3 or 6 months

519
00:20:18,894 --> 00:20:20,595
throughout the PhD journey.

520
00:20:20,974 --> 00:20:23,214
And it turns out that every single time,

521
00:20:23,214 --> 00:20:24,974
no matter how ambitious we are at the

522
00:20:24,974 --> 00:20:25,474
start,

523
00:20:25,934 --> 00:20:28,015
at the end, the students will have done

524
00:20:28,015 --> 00:20:30,335
so much more than they ever thought possible

525
00:20:30,335 --> 00:20:31,075
at the beginning.

526
00:20:31,375 --> 00:20:33,775
So it's a very dynamic approach we take

527
00:20:33,775 --> 00:20:34,275
to

528
00:20:34,710 --> 00:20:36,869
supervision. Of course, we also have the advantage

529
00:20:36,869 --> 00:20:39,509
of having several cohorts of students and quite

530
00:20:39,509 --> 00:20:41,109
a large number of students, and we bring

531
00:20:41,109 --> 00:20:43,509
them together on a regular basis so that

532
00:20:43,509 --> 00:20:45,509
they have a chance to compare. So you

533
00:20:45,509 --> 00:20:46,009
mentioned,

534
00:20:46,789 --> 00:20:48,884
the the different structures that exist

535
00:20:49,204 --> 00:20:50,825
between supervisors acting

536
00:20:51,285 --> 00:20:53,924
more like, a line manager, say, or or

537
00:20:53,924 --> 00:20:56,505
being a colleague in in the research laboratory.

538
00:20:57,045 --> 00:21:00,164
And, students can compare between themselves what,

539
00:21:00,644 --> 00:21:03,144
what the different approaches are, how these work.

540
00:21:03,319 --> 00:21:05,740
And then in a dialogue with their supervisory

541
00:21:05,960 --> 00:21:08,920
team, they can shape and reshape the way

542
00:21:08,920 --> 00:21:11,400
that their own PhD project is working. And

543
00:21:11,400 --> 00:21:13,980
in LIF Inno, this has worked incredibly well.

544
00:21:14,200 --> 00:21:15,339
So as you were talking

545
00:21:15,880 --> 00:21:16,619
there, Carsten,

546
00:21:17,079 --> 00:21:18,619
what I was thinking of was

547
00:21:20,105 --> 00:21:22,264
I I can imagine here from what I

548
00:21:22,264 --> 00:21:24,764
remember of of working as a PhD student

549
00:21:24,825 --> 00:21:25,325
that

550
00:21:25,784 --> 00:21:29,404
sometimes the idea of PhD students doing placements

551
00:21:30,424 --> 00:21:32,845
and having additional training that is not

552
00:21:33,890 --> 00:21:35,509
immediately geared towards

553
00:21:35,890 --> 00:21:38,690
learning new science for their PhD might get

554
00:21:38,690 --> 00:21:39,190
some

555
00:21:40,210 --> 00:21:42,450
pushback from supervisors who might think, well, why

556
00:21:42,450 --> 00:21:44,130
is my student, you know, spending all this

557
00:21:44,130 --> 00:21:45,890
time when they should be working for me

558
00:21:45,890 --> 00:21:46,390
instead

559
00:21:47,244 --> 00:21:49,724
doing, you know, doing this training that isn't

560
00:21:49,724 --> 00:21:51,964
going to directly benefit my group. So actually,

561
00:21:51,964 --> 00:21:53,804
I'm going to kick this question over to

562
00:21:53,804 --> 00:21:54,304
Andrea.

563
00:21:55,565 --> 00:21:56,944
What do you think are the benefits

564
00:21:57,565 --> 00:21:58,065
for,

565
00:21:59,680 --> 00:22:00,420
for academics

566
00:22:00,799 --> 00:22:03,519
specifically of having students who are working on

567
00:22:03,519 --> 00:22:04,580
these kinds of CBTs?

568
00:22:08,480 --> 00:22:11,599
Well, there's benefits for supervisors, but also for

569
00:22:11,599 --> 00:22:13,859
students as well. And I think it's benefits

570
00:22:13,994 --> 00:22:16,234
for the projects in general. So what I

571
00:22:16,234 --> 00:22:19,055
notice is people, students who are actually

572
00:22:19,595 --> 00:22:22,394
going to placements, they they come back more

573
00:22:22,394 --> 00:22:22,894
energized.

574
00:22:23,515 --> 00:22:24,575
They do have,

575
00:22:25,595 --> 00:22:26,734
they they learn,

576
00:22:27,434 --> 00:22:30,059
how to work better in teams, how to

577
00:22:30,059 --> 00:22:30,559
deliver

578
00:22:31,980 --> 00:22:34,640
projects on shorter time skills that are usually

579
00:22:34,779 --> 00:22:35,759
done in academia.

580
00:22:37,099 --> 00:22:38,799
They come back with ideas

581
00:22:39,579 --> 00:22:40,799
about how to

582
00:22:41,259 --> 00:22:43,200
extend their research and make

583
00:22:43,579 --> 00:22:46,214
it great spin offs or have a more

584
00:22:46,214 --> 00:22:46,714
entrepreneurial

585
00:22:47,335 --> 00:22:47,835
mindset,

586
00:22:49,575 --> 00:22:51,815
in general, just have a broader perspective. So

587
00:22:51,815 --> 00:22:54,315
I think that's very useful for both students

588
00:22:54,375 --> 00:22:55,595
and and supervisors.

589
00:22:56,375 --> 00:22:57,990
So every student is

590
00:22:59,509 --> 00:23:01,529
doing a placement of a minimum,

591
00:23:01,910 --> 00:23:03,529
duration of 6 months,

592
00:23:04,549 --> 00:23:07,929
and but some placements take longer depending on

593
00:23:08,150 --> 00:23:09,609
the direction of the project.

594
00:23:10,224 --> 00:23:11,524
We have some cases

595
00:23:11,825 --> 00:23:15,365
where, our industry partners are jointly supervising,

596
00:23:16,464 --> 00:23:18,644
with with the research partners

597
00:23:19,024 --> 00:23:19,764
so that,

598
00:23:20,544 --> 00:23:22,164
that the placement actually,

599
00:23:22,625 --> 00:23:25,740
becomes our integral part of the PhD thesis.

600
00:23:27,420 --> 00:23:29,019
Vikosti, can you give an example of one

601
00:23:29,019 --> 00:23:30,799
of the projects that is from the start

602
00:23:31,420 --> 00:23:33,440
partnered within with an industry partner?

603
00:23:34,460 --> 00:23:37,420
Yeah. Maybe maybe just one more word about

604
00:23:37,420 --> 00:23:40,355
what you said earlier in terms of pushback

605
00:23:40,494 --> 00:23:40,994
coming

606
00:23:41,454 --> 00:23:43,375
from the students. I think it's fair to

607
00:23:43,375 --> 00:23:44,434
say that when

608
00:23:44,815 --> 00:23:45,315
we

609
00:23:45,775 --> 00:23:47,554
first established the idea

610
00:23:48,174 --> 00:23:49,954
of having a 6 months placement

611
00:23:50,575 --> 00:23:53,075
integral to all of the PhD projects,

612
00:23:53,619 --> 00:23:55,940
there was pushback from everybody. From the students,

613
00:23:55,940 --> 00:23:56,919
from the supervisory

614
00:23:57,220 --> 00:23:59,220
team, who, as you said, didn't want to

615
00:23:59,220 --> 00:23:59,880
see their

616
00:24:00,259 --> 00:24:02,200
student waste time

617
00:24:02,500 --> 00:24:03,000
on,

618
00:24:03,619 --> 00:24:06,179
research that isn't directly relevant for the PhD,

619
00:24:06,179 --> 00:24:08,740
but also from the industry partners because they

620
00:24:08,740 --> 00:24:10,014
felt that

621
00:24:10,315 --> 00:24:10,894
they needed,

622
00:24:11,914 --> 00:24:14,335
focused staff rather than students

623
00:24:14,714 --> 00:24:17,054
working on their data related challenges.

624
00:24:17,674 --> 00:24:18,154
Now,

625
00:24:18,954 --> 00:24:20,554
that means that at the beginning, we did

626
00:24:20,554 --> 00:24:22,014
have a hard time to communicate

627
00:24:22,669 --> 00:24:24,990
that requirement to everybody and to get buy

628
00:24:24,990 --> 00:24:27,069
into the idea, and I think what really

629
00:24:27,069 --> 00:24:27,569
changed

630
00:24:28,029 --> 00:24:30,829
everyone's minds was the great success that these

631
00:24:30,829 --> 00:24:33,390
placements have delivered. When the students came back,

632
00:24:33,390 --> 00:24:36,164
as Andrea said, they became back more energized.

633
00:24:36,464 --> 00:24:38,865
The academics all of a sudden had industry

634
00:24:38,865 --> 00:24:41,125
links that they would never have had before.

635
00:24:41,424 --> 00:24:44,464
Academics are, most of them, not incredibly good

636
00:24:44,464 --> 00:24:47,105
at establishing new links with industry partners they

637
00:24:47,105 --> 00:24:49,559
have never met before, and our center just

638
00:24:49,880 --> 00:24:52,840
provides these industry links automatically, so they just

639
00:24:52,840 --> 00:24:55,480
have to basically be there and engage in

640
00:24:55,480 --> 00:24:55,980
that,

641
00:24:56,440 --> 00:24:59,080
constructive partnership. And industry all of a sudden

642
00:24:59,080 --> 00:25:00,779
saw that these students,

643
00:25:01,160 --> 00:25:03,744
they are actually in the 3rd or final

644
00:25:03,744 --> 00:25:05,744
year of their PhD, so they are very,

645
00:25:05,744 --> 00:25:06,964
very highly skilled,

646
00:25:07,265 --> 00:25:09,744
and they can help them very efficiently to

647
00:25:09,744 --> 00:25:10,565
solve problems.

648
00:25:10,865 --> 00:25:13,605
At the same time, for industry, these placements

649
00:25:13,664 --> 00:25:16,029
have also become a recruitment tool because they

650
00:25:16,029 --> 00:25:18,430
have an opportunity to test somebody for 6

651
00:25:18,430 --> 00:25:19,329
months continuously.

652
00:25:19,869 --> 00:25:21,710
And if they like what they see, they

653
00:25:21,710 --> 00:25:23,309
make them an offer. And this has happened

654
00:25:23,309 --> 00:25:25,250
now several times that the students

655
00:25:25,549 --> 00:25:28,029
that conducted placements were made a job offer

656
00:25:28,029 --> 00:25:29,630
in the very company that they did the

657
00:25:29,630 --> 00:25:32,115
placement in. And these were all companies that

658
00:25:32,115 --> 00:25:33,815
they didn't have on their career

659
00:25:34,275 --> 00:25:37,235
radar beforehand. They were companies that they wanted

660
00:25:37,235 --> 00:25:38,835
to explore, and all of a sudden it

661
00:25:38,835 --> 00:25:40,215
became so much more.

662
00:25:40,595 --> 00:25:42,434
So I guess the question is then how

663
00:25:42,434 --> 00:25:44,880
do we approach these companies? How do we

664
00:25:45,119 --> 00:25:47,279
find the ones that we offer to our

665
00:25:47,279 --> 00:25:50,079
students? And, again, that comes through dialogue with

666
00:25:50,079 --> 00:25:52,559
the students. Firstly, we ask the students whether

667
00:25:52,559 --> 00:25:54,099
there are maybe some companies

668
00:25:54,400 --> 00:25:56,400
they would like to do a placement with.

669
00:25:56,400 --> 00:25:58,319
And sometimes students would like to work with

670
00:25:58,319 --> 00:26:00,019
a global player, say IBM,

671
00:26:00,514 --> 00:26:02,835
Microsoft, Google to just name a few, and

672
00:26:02,835 --> 00:26:05,095
then we create a link to these companies

673
00:26:05,154 --> 00:26:07,954
and see what opportunities there are. Sometimes they

674
00:26:07,954 --> 00:26:08,454
also

675
00:26:08,755 --> 00:26:11,075
would like to explore a much smaller company

676
00:26:11,075 --> 00:26:12,835
in the region because they see that as

677
00:26:12,835 --> 00:26:14,615
a potential employment opportunity.

678
00:26:15,000 --> 00:26:17,400
And then, again, we do help to create

679
00:26:17,400 --> 00:26:18,059
that link.

680
00:26:18,440 --> 00:26:20,519
And, we do, of course, also have a

681
00:26:20,519 --> 00:26:22,940
very long list of companies that is now

682
00:26:23,160 --> 00:26:25,400
connected with our CDT where we can also

683
00:26:25,400 --> 00:26:27,880
make proposals of what we think is a

684
00:26:27,880 --> 00:26:30,544
good fit to our students' interest. Now if

685
00:26:30,544 --> 00:26:32,464
I can give one specific example of a

686
00:26:32,464 --> 00:26:34,704
journey that we had with a company, it's

687
00:26:34,704 --> 00:26:36,244
a relatively small company,

688
00:26:36,785 --> 00:26:37,765
called Adaptics,

689
00:26:38,464 --> 00:26:41,424
an SME that is specialized in 3 d

690
00:26:41,424 --> 00:26:42,484
imaging technologies.

691
00:26:42,980 --> 00:26:45,460
And we started working with them maybe 10

692
00:26:45,460 --> 00:26:47,299
years ago. And at the beginning, it really

693
00:26:47,299 --> 00:26:47,799
was,

694
00:26:48,580 --> 00:26:49,480
a a very,

695
00:26:50,099 --> 00:26:52,599
simple partnership. We had some computational

696
00:26:52,900 --> 00:26:55,380
and modeling skills that we thought would be

697
00:26:55,380 --> 00:26:56,565
beneficial for them,

698
00:26:56,964 --> 00:27:00,244
fundamentally, using Monte Carlo based approaches to model

699
00:27:00,244 --> 00:27:01,464
radiation impacting

700
00:27:01,924 --> 00:27:03,764
on the skull of a patient and then

701
00:27:03,764 --> 00:27:04,264
seeing,

702
00:27:04,804 --> 00:27:07,365
what kind of intensity distributions we get, how

703
00:27:07,365 --> 00:27:10,004
we can understand errors and error propagation and

704
00:27:10,004 --> 00:27:12,309
these kind of things. Now that has developed

705
00:27:12,769 --> 00:27:16,150
fantastically over that period. It started with just,

706
00:27:16,609 --> 00:27:18,369
a little bit of overlap. We then had

707
00:27:18,369 --> 00:27:20,769
a first PhD student, which was funded by

708
00:27:20,769 --> 00:27:21,430
the university

709
00:27:22,049 --> 00:27:24,390
working on an industry related,

710
00:27:25,009 --> 00:27:28,205
problem that Adaptics had. Then we continued the

711
00:27:28,205 --> 00:27:31,005
journey in Livinnow by now having a student

712
00:27:31,005 --> 00:27:32,705
who is jointly funded 5050

713
00:27:33,245 --> 00:27:34,384
between the company

714
00:27:34,684 --> 00:27:35,505
and the center.

715
00:27:35,884 --> 00:27:37,105
And very recently,

716
00:27:37,485 --> 00:27:39,085
in fact, just a few weeks ago, we

717
00:27:39,085 --> 00:27:41,059
were now awarded funding by SDFC

718
00:27:41,359 --> 00:27:43,679
for a half a million pound project that

719
00:27:43,679 --> 00:27:45,859
will take the technology that we have jointly

720
00:27:45,919 --> 00:27:48,259
developed to the next level. So it's really

721
00:27:48,319 --> 00:27:50,500
a success story that shows that we

722
00:27:51,119 --> 00:27:53,220
are prepared to engage with industry

723
00:27:53,695 --> 00:27:55,615
at a very small level initially. We want

724
00:27:55,615 --> 00:27:57,375
to show that we are serious about this.

725
00:27:57,375 --> 00:27:59,695
We want to show that the students can

726
00:27:59,695 --> 00:28:02,734
really work extremely well with that business. And

727
00:28:02,734 --> 00:28:04,894
then if if all the sides see the

728
00:28:04,894 --> 00:28:05,394
advantages

729
00:28:05,855 --> 00:28:08,329
that are in it for them, then automatically,

730
00:28:08,630 --> 00:28:10,410
almost, this will grow over time.

731
00:28:11,109 --> 00:28:13,109
And as you mentioned, the CEC has been

732
00:28:13,109 --> 00:28:13,609
running

733
00:28:14,390 --> 00:28:15,690
in in some iteration

734
00:28:15,990 --> 00:28:17,669
for for a while now. When it when

735
00:28:17,669 --> 00:28:19,289
you first started, was it difficult

736
00:28:19,795 --> 00:28:20,615
to approach,

737
00:28:21,075 --> 00:28:23,075
industry and sort of make the case that

738
00:28:23,075 --> 00:28:24,295
these companies should have,

739
00:28:25,795 --> 00:28:28,454
a student work with them for 6 months?

740
00:28:28,835 --> 00:28:31,154
Yeah. It was very difficult, and we had

741
00:28:31,154 --> 00:28:33,474
to spend considerable time. I had personally to

742
00:28:33,474 --> 00:28:34,934
spend considerable time,

743
00:28:35,589 --> 00:28:38,789
talking with businesses, explaining the skills our students

744
00:28:38,789 --> 00:28:40,169
had, explaining why

745
00:28:40,470 --> 00:28:42,549
I thought, why we thought that it would

746
00:28:42,549 --> 00:28:44,809
be beneficial to put them onto their

747
00:28:45,190 --> 00:28:47,049
ongoing research projects.

748
00:28:47,444 --> 00:28:50,404
And, what really then changed the mindset of,

749
00:28:51,284 --> 00:28:52,345
companies subsequently

750
00:28:53,125 --> 00:28:55,125
were the success stories that we had from

751
00:28:55,125 --> 00:28:55,944
these initial,

752
00:28:56,644 --> 00:28:59,365
companies and the fantastic results that came out

753
00:28:59,365 --> 00:29:00,184
of our research.

754
00:29:01,220 --> 00:29:03,160
Now can you talk a bit about what,

755
00:29:03,859 --> 00:29:05,720
what's in store for the future of,

756
00:29:06,500 --> 00:29:07,400
of the CBC?

757
00:29:08,500 --> 00:29:10,180
Yeah. I think it's fair to say that

758
00:29:10,180 --> 00:29:12,259
we have come a long way over the

759
00:29:12,259 --> 00:29:14,585
last 10 years. And of course we would

760
00:29:14,585 --> 00:29:17,065
like to continue our successful journey and and

761
00:29:17,065 --> 00:29:19,644
offer more students the opportunity really

762
00:29:19,945 --> 00:29:22,424
to be trained in the LiveInno way as

763
00:29:22,424 --> 00:29:23,484
you might call it.

764
00:29:23,865 --> 00:29:25,545
So we we do hope that we can

765
00:29:25,545 --> 00:29:29,160
get additional studentships funded from STFC and other

766
00:29:29,160 --> 00:29:32,920
funders but also from external partners, research centers

767
00:29:32,920 --> 00:29:35,559
and industry. We've really seen that this way

768
00:29:35,559 --> 00:29:38,775
of training students has worked incredibly well

769
00:29:39,174 --> 00:29:41,494
so we really would like to continue that

770
00:29:41,494 --> 00:29:44,394
journey with our partners. In terms of research,

771
00:29:44,775 --> 00:29:46,695
I think what we've seen in the past

772
00:29:46,695 --> 00:29:49,255
2 years is an ever stronger focus on

773
00:29:49,255 --> 00:29:49,755
AI,

774
00:29:50,455 --> 00:29:53,174
which is becoming increasingly important for all of

775
00:29:53,174 --> 00:29:54,555
our research areas.

776
00:29:55,150 --> 00:29:57,009
And, I think we are well positioned

777
00:29:57,390 --> 00:30:00,430
for tackling the research challenges in this area

778
00:30:00,430 --> 00:30:02,529
across all of our different communities.

779
00:30:03,470 --> 00:30:05,890
For example, we've just recruited a new lecturer

780
00:30:06,190 --> 00:30:08,349
at the University of Liverpool who now has

781
00:30:08,349 --> 00:30:09,970
a focus on AI for

782
00:30:10,275 --> 00:30:12,855
particle accelerator. So also in terms of the

783
00:30:12,914 --> 00:30:14,134
academic talent,

784
00:30:14,595 --> 00:30:16,755
we have really positioned us well for the

785
00:30:16,755 --> 00:30:17,974
challenges of the future.

786
00:30:18,755 --> 00:30:19,894
And are there any,

787
00:30:20,515 --> 00:30:21,575
I guess, upcoming

788
00:30:22,115 --> 00:30:23,634
if someone is listening to this and that

789
00:30:23,634 --> 00:30:24,869
either, you know, someone who

790
00:30:25,430 --> 00:30:28,329
works in industries interested in getting involved or

791
00:30:28,549 --> 00:30:29,049
is,

792
00:30:29,589 --> 00:30:30,490
you know, say,

793
00:30:31,269 --> 00:30:33,109
an undergraduate student who thinks this sounds like

794
00:30:33,109 --> 00:30:35,210
something they're interested in, are there any upcoming,

795
00:30:36,789 --> 00:30:38,650
deadlines that they should be aware of?

796
00:30:39,325 --> 00:30:41,805
Yeah. So every year, we have an open

797
00:30:41,805 --> 00:30:44,625
call to academics in the departments of physics,

798
00:30:44,684 --> 00:30:47,984
mathematics, computer science, engineering, and we ask them

799
00:30:48,045 --> 00:30:50,845
for their data science PhD projects. And then

800
00:30:50,845 --> 00:30:51,825
there is a prioritization

801
00:30:52,205 --> 00:30:52,705
process.

802
00:30:53,099 --> 00:30:55,339
And typically at the end of October, early

803
00:30:55,339 --> 00:30:56,480
November, we publish

804
00:30:56,779 --> 00:30:59,259
these projects on our website, and then there's

805
00:30:59,259 --> 00:31:01,420
a first application deadline at the end of

806
00:31:01,420 --> 00:31:03,900
January. So I would encourage everyone who is

807
00:31:03,900 --> 00:31:07,494
interested in doing a PhD within LIFINTO to

808
00:31:07,515 --> 00:31:09,755
have a careful look at our website when

809
00:31:09,755 --> 00:31:12,174
these projects are released later this year.

810
00:31:12,875 --> 00:31:15,035
The first version of ChatCBT, that would have

811
00:31:15,035 --> 00:31:17,054
come out in oh, it's 2021.

812
00:31:17,595 --> 00:31:19,194
So this would have been yeah. Obviously, while

813
00:31:19,194 --> 00:31:21,490
the CDC was running, did you did you

814
00:31:21,490 --> 00:31:23,490
clock that when it happened? Because for me,

815
00:31:23,490 --> 00:31:24,549
that's sort of the

816
00:31:25,169 --> 00:31:27,029
the start of these, like,

817
00:31:28,289 --> 00:31:29,730
models that I think actually a lot of

818
00:31:29,730 --> 00:31:31,990
people use for their in their research now

819
00:31:32,049 --> 00:31:33,809
becoming a really big thing. Did you did

820
00:31:33,809 --> 00:31:35,089
you clock that and you think, oh, this

821
00:31:35,089 --> 00:31:37,045
is something that is gonna really change the

822
00:31:37,045 --> 00:31:38,424
way that our students work.

823
00:31:38,725 --> 00:31:41,205
We we did, in 2 different ways. On

824
00:31:41,205 --> 00:31:43,125
the one hand, of course, we all of

825
00:31:43,125 --> 00:31:45,465
a sudden needed to really put some emphasis

826
00:31:45,605 --> 00:31:46,105
on

827
00:31:47,365 --> 00:31:49,785
teaching our students about prompt engineering,

828
00:31:50,950 --> 00:31:52,869
so so how to really use those tools

829
00:31:52,869 --> 00:31:53,369
efficiently.

830
00:31:54,070 --> 00:31:54,809
But maybe

831
00:31:55,110 --> 00:31:56,809
and maybe even more importantly,

832
00:31:57,670 --> 00:31:59,750
we also have to teach them about the

833
00:31:59,750 --> 00:32:00,250
limitations,

834
00:32:00,549 --> 00:32:03,210
very fundamental limitations of these tools,

835
00:32:03,830 --> 00:32:05,210
because if wrongly

836
00:32:05,884 --> 00:32:07,424
applied or applied for

837
00:32:07,804 --> 00:32:11,184
the wrong questions, they can be completely wrong.

838
00:32:11,325 --> 00:32:13,085
And I think today is one of the

839
00:32:13,085 --> 00:32:15,184
big challenges is to really distinguish,

840
00:32:16,044 --> 00:32:17,105
between fact,

841
00:32:17,565 --> 00:32:18,065
and

842
00:32:18,524 --> 00:32:19,024
generated

843
00:32:19,484 --> 00:32:20,304
soil knowledge,

844
00:32:21,299 --> 00:32:24,339
where tools suggest that they give you a

845
00:32:24,339 --> 00:32:26,899
very nice and rounded text, but where if

846
00:32:26,899 --> 00:32:29,079
you look in the details, which is particularly

847
00:32:29,220 --> 00:32:29,720
relevant,

848
00:32:30,500 --> 00:32:31,559
in in research,

849
00:32:32,259 --> 00:32:35,684
they are quite fundamentally wrong. And learning this,

850
00:32:36,085 --> 00:32:37,924
learning how to deal with that and learning

851
00:32:37,924 --> 00:32:39,625
where to find quality information

852
00:32:40,325 --> 00:32:41,545
is really something,

853
00:32:42,244 --> 00:32:44,884
that gets stronger a stronger and stronger focus

854
00:32:44,884 --> 00:32:45,785
in our training.

855
00:32:46,420 --> 00:32:47,940
So the students on the CVC at the

856
00:32:47,940 --> 00:32:49,960
moment, they're gonna finish in 2028.

857
00:32:50,339 --> 00:32:52,500
Right? Yep. That's correct. Right. And I can

858
00:32:52,500 --> 00:32:54,759
imagine that if you if you think of

859
00:32:54,980 --> 00:32:57,859
how quickly, for example, something like ChatCBT has

860
00:32:57,859 --> 00:32:59,799
evolved, I can imagine that

861
00:33:00,420 --> 00:33:01,799
by the time they finish,

862
00:33:02,875 --> 00:33:05,674
the landscape around AI and machine learning is

863
00:33:05,674 --> 00:33:06,174
gonna

864
00:33:07,595 --> 00:33:09,674
look quite different. Are you kind of keeping

865
00:33:09,674 --> 00:33:11,355
an eye on that? Are you anticipating that

866
00:33:11,355 --> 00:33:12,654
you might have to sort of

867
00:33:13,755 --> 00:33:16,750
maybe adapt the CDT over the next couple

868
00:33:16,750 --> 00:33:18,289
of years as that stuff changes?

869
00:33:18,990 --> 00:33:20,910
Yeah. We definitely have to adapt.

870
00:33:21,309 --> 00:33:23,809
And as you you're right, we say AI,

871
00:33:24,430 --> 00:33:26,049
driven tools like ChatGPT

872
00:33:26,670 --> 00:33:28,990
are becoming increasingly important. And I guess that's

873
00:33:28,990 --> 00:33:30,690
one of the strengths of our

874
00:33:31,035 --> 00:33:34,575
approach of using that structured career development plan

875
00:33:35,115 --> 00:33:37,775
where, by definition, every 3 to 6 months,

876
00:33:37,994 --> 00:33:40,555
the supervisory team sits down with the student,

877
00:33:40,555 --> 00:33:42,315
and they look at every aspect of the

878
00:33:42,315 --> 00:33:43,295
project, including

879
00:33:43,835 --> 00:33:46,315
the training and the skills that the student

880
00:33:46,315 --> 00:33:49,539
needs. And then if centrally we realize that

881
00:33:49,539 --> 00:33:51,140
there is now a big push in a

882
00:33:51,140 --> 00:33:53,319
new area that simply didn't exist,

883
00:33:54,019 --> 00:33:55,940
a few years ago, then we build this

884
00:33:55,940 --> 00:33:58,200
into the training program. We either

885
00:33:58,500 --> 00:34:00,660
train our own experts or we work with

886
00:34:00,660 --> 00:34:01,720
external specialists,

887
00:34:02,025 --> 00:34:03,865
and then we bring those skills to the

888
00:34:03,865 --> 00:34:06,424
students and make sure they are positioned best

889
00:34:06,424 --> 00:34:07,644
for their future careers.

890
00:34:09,144 --> 00:34:10,744
Thank you so much for joining me. You've

891
00:34:10,744 --> 00:34:11,784
been listening to,

892
00:34:12,105 --> 00:34:15,079
Carson Walsh and Andrea Fonte of the LIV

893
00:34:15,079 --> 00:34:17,579
Inno Centre For Doctoral Training in Liverpool.

894
00:34:18,679 --> 00:34:19,739
Thank you so much.

895
00:34:26,599 --> 00:34:28,280
I'm afraid that's all the time we have

896
00:34:28,280 --> 00:34:29,500
for this week's podcast.

897
00:34:29,985 --> 00:34:31,684
Thanks to Andrea Fonte,

898
00:34:31,985 --> 00:34:35,605
Carsten Welsh, and Katherine Skipper for a fascinating

899
00:34:35,905 --> 00:34:36,405
conversation,

900
00:34:36,945 --> 00:34:39,585
and a special thanks to our producer, Fred

901
00:34:39,585 --> 00:34:40,085
Ailes.

902
00:34:40,704 --> 00:34:43,204
This episode is sponsored by LiveInno,

903
00:34:43,989 --> 00:34:47,610
the Liverpool Centre For Doctoral Training For Innovation

904
00:34:47,989 --> 00:34:50,010
in Data Intensive Science.

905
00:34:50,550 --> 00:34:52,410
The UK based CDT

906
00:34:52,869 --> 00:34:55,530
is a partnership between the University of Liverpool

907
00:34:55,829 --> 00:34:58,170
and Liverpool John Moores University.

908
00:34:58,994 --> 00:34:59,894
Since 2022,

909
00:35:00,835 --> 00:35:03,255
LiveInno has offered 3 cohorts

910
00:35:03,554 --> 00:35:04,295
of students

911
00:35:04,674 --> 00:35:06,454
4 year fully funded

912
00:35:06,755 --> 00:35:07,255
PhD

913
00:35:07,634 --> 00:35:08,134
studentships

914
00:35:08,594 --> 00:35:10,534
in data intensive science.

915
00:35:11,234 --> 00:35:13,734
From medical physics to quantum computing,

916
00:35:14,049 --> 00:35:16,710
the CDT has a diverse portfolio

917
00:35:17,170 --> 00:35:18,549
of research projects,

918
00:35:18,849 --> 00:35:22,150
but students all share training in high performance

919
00:35:22,210 --> 00:35:22,710
computing,

920
00:35:23,089 --> 00:35:26,710
machine learning, and AI, and data analysis.

921
00:35:27,425 --> 00:35:28,085
The CDT

922
00:35:28,465 --> 00:35:31,744
also gives students access to career advice and

923
00:35:31,744 --> 00:35:33,764
training in project management,

924
00:35:34,304 --> 00:35:34,804
entrepreneurship,

925
00:35:35,425 --> 00:35:36,164
and communication

926
00:35:36,465 --> 00:35:36,965
skills.

927
00:35:37,505 --> 00:35:40,085
As well as working on their original research,

928
00:35:40,530 --> 00:35:42,710
each student undertakes a 6 month

929
00:35:43,010 --> 00:35:46,610
industrial placement, and some projects are sponsored by

930
00:35:46,610 --> 00:35:48,789
an industrial partner throughout.

931
00:35:49,570 --> 00:35:52,769
The next cohort of PhD projects with a

932
00:35:52,769 --> 00:35:54,230
start date of autumn

933
00:35:54,695 --> 00:35:55,195
2025

934
00:35:55,894 --> 00:35:57,195
will be announced soon.

935
00:35:57,574 --> 00:36:00,235
More details can be found at www.liveinno.org.