Testing the water
Industry placements for PhD students can be hugely beneficial for all concerned. We hear from two former students and the companies they worked for.
Dr Lars Henklemann & OKRA.ai, an Envision Pharma Group company
Why Cambridge? Before I came to Cambridge to do a PhD, I was an undergraduate at Heidelberg University studying particle physics. My supervisor there had accepted me on to a PhD programme but was then offered a job in Cambridge. I was able to transfer my offer to Cambridge and to get the funding I needed, I applied to the Centre for Doctoral Training (CDT) in Data Intensive Science.
CDTs include an industry placement as part of the training they offer. Was that a particular attraction? At that stage, I was probably just thinking about the best way to get funding for my PhD. But it was nice to have half a year to test the water and see what it's like to work in industry even though I didn't think I would stick with it.
Your PhD in a nutshell? I was working with the ATLAS Detector at the Large Hadron Collider at CERN, recording and analysing what happens when protons collide, looking for a new particle hypothesised by physicists in their proposed explanations for dark matter.
How did the placement with OKRA.ai come about? I had heard good things about the company from previous students and I liked the fact that it's enabling healthcare. The other thing I like is that the industry is very regulated. This means that when we use AI we already have to comply with strict regulations which other AI sectors are still catching up on.
What did you do during your placement? OKRA has a product called the 'Medical Brain'. It connects huge volumes of text, from healthcare providers, researchers and pharma companies. Bringing them together produces tons of interesting insights and by using machine learning we are able to identify emerging topics and trends. I was looking at how to integrate large language models into the 'Brain'.
Once you've found an interesting topic, you can present the raw data relating to it. But that can be very messy and difficult to get into. It's much better if you can generate a nice clear summary that tells you what's going on with links to the underlying data. Creating these user-friendly summaries was one aspect of my work; the other was to develop a chatbot to help people access the insights without needing to use the dashe.
What was the company like to work for? I felt very well supported from day one. They have team lunches every Wednesday. Not only is the food delicious but it's also fun because you meet people that you don't get to talk to as much.
Top three things you learnt from the placement?
1. A new world view!
Prior to OKRA.ai I had never used natural language processing. I'm used to using numbers to describe the physical world but I've always thought of language as this vague, human, messy thing. I've realised that, in spite of its vagueness, by using mathematical techniques you can go a very long way to pinning language down. In some ways, that realisation has changed how I look at the world.
2. Good data housekeeping
I was really impressed with the amount of care taken to keep track of the data at OKRA.ai. We aim to be like that in academia but it's not always as well implemented in practice. If I had gone back to research, I would have wanted to put that knowledge to good use.
3. Collaboration
In the ATLAS collaboration there are about 3,000 people who you somehow have to align with. It's much easier to work with small numbers of people in the same building. Even when there's strong disagreement to start with, if you can get people in a room together, you can find a way through it.
You are now working for OKRA.ai. When did you decide this was the right path for you? I realised I wanted to continue developing the prototypes I'd been working on during the placement. And I just really enjoyed the day-to-day work and was excited to carry on with it.
What's your role now? I'm a data scientist in the natural language processing team. A lot of what I'm doing is a continuation of what I started as an intern but there's also a lot of aligning with people about what we are going to do, why and how. I'm really enjoying it!
The company view
Alice Di Giulio, AI Policy & Public Affairs Lead at OKRA.ai (an Envision Pharma Group company)
Why does OKRA.ai have student placements? We want to bring in expertise from the cutting edge of academia and different perspectives.
Recently, the University of Cambridge and Envision Pharma Group hosted a hackathon on the use AI for global healthcare, bringing together students and industry experts to solve real world cases with AI. We look forward to many more collaborations with Cambridge and connecting their bright minds to the life sciences industry to deploy their technical skills to drive faster and smarter patient outcomes.
What about Lars in particular? We've never had someone come in on a placement and drive their own work, in the way Lars has done. He's one of those people who you can throw any problem at and he'll find a solution.
With his expertise and contagious laugh, Lars is an invaluable member of our diverse team.
Dr Jonathan Itcovitz & SATAVIA
Why Cambridge? As an undergraduate, I studied physics at Imperial College London. Towards the end of my degree I specialised in topics to do with atmospheric processes.
But I've always loved space and particularly the physics of the solar system. I saw an opportunity at the Institute of Astronomy in Cambridge for a summer internship. I did that for a few months and really loved both the department and the city. So, when the time came, I applied for a PhD position there and was fortunate enough to get a place.
How did your industry placement come about? Cambridge allows PhD students to take up to six months out from their studies, suspending their funding, to do things like industry placements. I was keen to experience this opportunity before the end of my studies, so that I could make a more informed decision on whether I wanted the next step of my career to be in academia or closer to industry.
Your PhD in a nutshell? Trying to better understand how the Earth’s atmosphere and surface were influenced by giant collisions with large (Mars-sized) asteroids during the Earth’s early lifetime.
So how did the placement with SATAVIA come about? They were suggested to me by Professor James Fergusson who looks after industry placements for the CDT and has a lot of contacts. I really wanted to use my physics and SATAVIA was perfect. It's an atmospheric science company, looking at mitigating the warming effect of aircraft-created condensation trails (contrails).
It was also great for me that it was a start-up. If you like a lot of structure and a defined role, maybe a larger company is better, but I like having a diverse set of tasks to do in my day-to-day work and I was able to get involved in lots of different things.
What were you doing for SATAVIA? SATAVIA has a weather prediction model to help aircraft avoid the conditions which cause contrail formation.
Contrail science is quite a specialised branch of atmospheric science using both weather prediction modelling and contrail modelling. You feed the weather parameters into our model and it creates a representation of what the contrail would look like if the aircraft was flying through that weather.
I was part of the science team, validating the model by running tests on it. I picked through all of the code to make sure it matched up with the literature and then used that knowledge to write up documentation to help with the business side of the company, such as for investors or partners within the aviation and climate industries.
Then there was the day-to-day operation. Satavia was actively involved in flight trials with its partner Etihad every week. I was there helping to make sure that everything made sense atmospherically.
Top three things you learnt from the placement?
1. Understanding what success looks like
How to set up your own frame of reference for what success is and to pace yourself into that. Breaking down large tasks into small achievable daily/weekly goals is really important.
2. Better coding
I'd written code for my own research but SATAVIA showed me how to do it in a much more professional way.
3. Communication skills
When you're doing scientific research, you are often speaking to people in your field or you're doing outreach to people who are in adjacent fields. I think that misses out a key category of people who are either in industry or are policymakers because they speak a very different language.
You have to learn how to how to communicate effectively with different types of people by trying to understand their motivations and what they will need as a result.
You've taken an academic post, back at Imperial. What is your role now? I'm a postdoc with my own research project, still working on modelling contrails and I'm still working with industry. I'm involved in a European consortium with other research institutes and industry partners looking at a large-scale solution for contrails.
The company view
Dr Conor Farrington, Chief of Staff at SATAVIA
Why does SATAVIA offer student placements? For their ideas! Someone in the later stages of their PhD will likely have exposure to very cutting-edge material, including research which has yet to be published.
What about Jonathan in particular? What we do is extremely complex, but Jonny was able to very quickly grasp the detail and represent it in slides to help us explain it clearly to lay audiences. It's very, very difficult to do that well.
We are currently developing a methodology to create incentives for airlines and others undertaking contrail management in routine aviation. Jonny was instrumental in helping us develop the scientific materials to support this approach, working at a very high level to advance the state of the art in this field.
Jonny's time with us gave us confidence that we could rely on the University’s student placement scheme to provide valuable and relevant expertise.
If your company would like to explore the options for hosting a PhD student, please get in touch via the Strategic Partnerships Office.
The text in this work is licensed under a Creative Commons Attribution 4.0 International License.
Published 29 May 2024
Photography
Banner image: Getty Images: Credit: Liudmila Chernetska
Images of Lars supplied by OKRA.ai
Images of Jonathan: credit: Jacqueline Garget
Built with Shorthand