Categories
Issue 8 Nonacademic Presenting Alumni

Alexandre Kempf: Blind shrimp, deaf mice, and how I found a job thanks to a YouTube video

🕒 19 min


Hi Alex, thank you for finding the time to join us for this interview. We are as happy to have you here in our Presenting Alumni session as we were to have you as a project leader in the last three editions of the S3 camp. Your career story is rather non-linear and filled with disappointing research outcomes, so I believe many young alumni will find it encouraging when they face similar things and decisions themselves. But first things first – how would you introduce yourself?

Alexandre Kempf, PhD
Data scientist at Pandascore, Paris

S3 2017, 2018 and 2019 project leader

Hi all, my name is Alexandre, though people usually call me Alex. I come from France, where I lived in Paris and Brittany during different stages of my life. I’m a computer scientist and a former neuroscientist, and I consider myself a bit geeky – I like to spend time writing computer code for fun or simply browsing websites to learn stuff. I feel like I can do a bunch of things, such as playing guitar or video games, at an average level, but am not super specialized in any single thing. Also, I’m a huge fan of pasta, cheese, and pesto, which practically make up my entire menu in these lockdown days!

Education: from bad to excellent grades & the story of disappointing research results ft. blind shrimp

What were you like as a child? Did you already take an interest in science in those early days?

I was an extremely calm, patient and easily influenced child. You could’ve left me sitting in a chair, given me something like a puzzle or a book, and when you returned five hours later you’d still find me in the same spot doing the same boring stuff. My big sister was kind of my role model and I’d do anything she told me to – as a consequence, I spent half of my childhood dressed as a girl. One time she managed to convince me to rub her feet every time there were commercials on TV, and while doing so I was actually thinking what a good deal I got (because it was only during the commercials, and not the entire time).

When it came to education, I was a hopelessly bad student during elementary school, despite all the help I got from my parents, who are a biology teacher and a mathematician. Once I got to the high school, something clicked and I became good at mathematics, which automatically made me an excellent student thanks to how the French education system works. I still can’t point my finger at how exactly that click happened; perhaps it was just a better math teacher.

How does that education system work exactly? I’d expect you’d need to master all the subjects to have an excellent overall grade, right?

Yes, that’s how a normal system would work. However, in France there are three different options to choose from and in each one the subjects are weighted differently. For example, in the option that I chose, math has a weight of 9, while the French language has a weight of only 1. So once I became good at math and other STEM subjects, everything else was less important. It was still a bit of a bumpy ride – at one point towards the end of high school, I somehow couldn’t grasp the new math concepts we were studying. Then during one class our teacher mentioned a math forum he was active on, so I went there to troll him, but ended up answering other people’s questions and actually catching up in my classes.

My biggest failure, though, was during the state exams at the end of high school. You know how my English is not exactly perfect? I know it’s difficult to believe, but my French is even worse. At the time I was really into the books of a famous French writer who perfected this way of writing in children’s voices. So there I am at the state exam in French, trying to imitate my idol writer, but doing it so poorly that exam correctors must’ve thought it was an actual child writing it, instead of a high school graduate. My score was a shameful 5/20, however a 19/20 in math saved me, because – remember – the weighted system.

Doing a chemistry experiment in high school.
  • High school: 2007 – 2010, Lycée Amiral Renarc’h, Brest
  • BSc in biology and mathematics: 2010 – 2012, Biology sea station, Roscoff
  • BSc in biology: 2012 – 2013, Ecole Normale Supérieure, Paris
  • Internship: 2013, National Museum of Natural History, Paris
  • Internship: 2014, Harvard University, Boston, USA
  • MSc in neuroscience, with minor degree in modelling and mathematics: 2013 – 2015, Ecole Normale Supérieure, Paris
  • PhD in neuroscience: 2015 – 2018, CNRS, Paris
  • Current: 2018 – , data scientist (computer vision and deep learning) at Pandascore, Paris

You know how they say – all’s well that ends well! How did you end up joining a bachelor’s program in biology and math?

My first idea was in fact to go to a medical school, following my best friend who I was also in love with, and to eventually become a neurosurgeon. After talking to my dad, I started to question this decision and ended up joining a 2-year-long program in biology and mathematics, which was novel at the time. For the final year of my bachelor’s, we were supposed to go somewhere outside of France, however I was really interested in the interdisciplinary program of Ecole Normale Supérieure in Paris, and was super lucky to have been accepted there because they only take the best students from France in their respective fields. And when I say interdisciplinary, that’s exactly what it was – even though my degree ended up being in biology, I was also able to go to classes such as history with the best history students in the country. It was an amazing experience.

That sounds quite unusual, but also really cool! Did you develop interests that will bring you to your internships while studying at Ecole Normale Supérieure?

A triop.

That’s exactly right! My favourite subject was neuroethology, which is a combination of neuroscience and behavioural science, the things I like the most (alongside evolution). I was never really into molecular biology – the stuff there is too tiny to be able to see it. I prefer looking at larger animals and how they behave and evolve. At some point, I had to give a presentation for this course, and the topic was zebra finches. I loved that topic so much that to this day I think it was the best presentation I’ve ever given.

Because of this obsession with songbirds, I tried to find a summer internship to do research on them, but I couldn’t find a thing. Eventually, my first internship at the National Museum of Natural History turned out to be on triops, which are these shrimp-like creatures with three brain cells, so not exactly smart animals. One of the guys who worked at the Museum kept them at home as pets and noticed that they sometimes make noises, so investigating those noises ended up being my research topic. I was basically placing microphones on top of their aquariums, only to learn that the mysterious noise was in fact blind triops bumping into aquarium glass. That was my first research failure – studying sounds of shrimpy animals that didn’t make sounds at all, but were in fact running into glass.

At least there’s now one less mystery to be solved in the animal world. Quite soon after, you actually did land the much wanted internship on songbirds in Boston. What was that like and in which way did that one “fail”?

The internship at Harvard during the first year of my master’s was really cool. I was supposed to study the impact of a small area of the brain on the learning of the zebra finches’ song. The research question there was: if we destroy this brain area, will the zebra finch learn how to sing? It won’t surprise you that the answer was yes, the birds will learn anyway.

Zebra finches (the most beautiful birds in Alex’s eyes).

Did this “failure” prompt you to give up researching songbirds?

Not by any means! While I was in Boston, I contacted a group in Paris that worked on zebra finches to work with them when I return to France. However, my supervisor at Harvard thankfully warned me not only that the professor in Paris is not a nice person to work with, but also that the zebra finches research will disappear because there’s not enough money in the field. He was himself in fact switching to studying rats.

So after searching a bit to figure out my next move, i.e. a topic for my master’s thesis, I found a group at CNRS using two-photon microscopy, which I found cool because it allows you to explore deeper layers of tissue, which is especially useful in neuroscience when you want to study the brain. The research there was on the interaction between vision and audition in mice cortices, and I was in fact doing it with a friend. So when we tested whether a visual input will affect each of these cortices, there was a lot happening in the visual cortex (my friend’s topic), and not a single thing in the auditory cortex (my topic, of course).

PhD and beyond: deaf mice, non-backed up data & a YouTube video that changed the course of my career

This is becoming quite the story of not giving up after facing disappointing research results. Your PhD research was also done on mice – I remember it was a running joke at S3 that you’re “playing music to deaf mice” for your PhD. How close to reality was that?

Well, they were partially deaf indeed. I was working with a mice strain commonly used in biology, which is known to lose sensitivity to high frequencies at the age of 5 – 8 months. What I typically did was perform a surgery when the mice were 4 months old – I’d usually remove part of their skull and replace it with a cranial window to be able to study the brain with a two-photon microscope, though sometimes I would also inject viruses or destroy parts of the brain. After the surgery, I would continue analysing the mouse for 1 – 2 months, when it would slowly go deaf – though I actually used only normal, not high frequencies.

Did you find it difficult to perform these surgeries, and to work with animals in general? I can’t picture myself doing such work.

That part of the PhD was the one I liked the least. The first time I had to kill a mouse after working with it for a few months, as well as the first time I failed during the surgery, I felt really bad. I’d say I was half-way to depression, thinking a lot about the ethics of that entire thing. There was some consolation in knowing that I wasn’t killing the mice right away, but was at least using them for a few months before I put them down. After a while I became detached, though, perhaps as a coping mechanism. I started to see mice more as a research subject than living things. The part I did enjoy was being able to see and touch this amazing organ – the brain. It made me feel almost superhuman.


What were the actual goals of your PhD research and what did the work look like day-to-day?

My PhD was on non-linearity of the auditory cortex in mice – if the neurons in mouse brains were linear, then increasing the sounds would cause an equal increase in response from the neurons; however, some neurons only respond to specific parts of sound, or only to quiet sounds. Basically everything that’s not linear was part of my research. The ultimate goal (which will be reached after two more decades of research, perhaps) is to help people with hearing impairments. At the moment, the hearing aid can only amplify sounds (linear boost). This makes it extremely impractical in noisy environments because all noises are amplified equally, so it’s impossible to ignore noises and focus on a conversation. Building a device that can recreate non-linearity in those circumstances would be a game-changer.

On a day-to-day basis, my first year mainly consisted of recording lots of microscopy data of mouse brains, while the remaining two years were spent on the analysis. However, at one point a guy in the lab deleted all my raw data.

That sounds immensely stressful! How come there was no data backup, and how did you end up finishing your PhD without the data?

So that was exactly the thing – there was supposed to be a backup. There is always a computer support person in the lab, however their contracts are very short-term. When a new person took that role, he mistakenly thought that the old one automated the data back up process, while in fact it was supposed to be done manually by running a script each week. I was relying on that system because data storage was still quite expensive in those days, especially for the huge amounts of data I was producing, and my lab didn’t have the money to invest into a separate backup system.

Fortunately, I did still have the compressed versions of my microscopy pictures, so I did manage to do some analysis and write my thesis. I couldn’t, however, go through with publishing a paper on the topic because I could easily imagine a scenario in which the reviewers would ask for additional analysis, which I wouldn’t be able to do without the raw data. In the end I finished my PhD with two papers on side-projects and a paper based on my master’s thesis work, but nothing on my main PhD project.

Successful PhD defence despite losing the research data!

Along the way, you kind of accidentally stumbled upon the place where you work now, which has nothing to do with neuroscience. How did that switch happen?

Once I lost my data, I became utterly bored and depressed. For a large part of that year, I spent my time on YouTube, and I really did go to the bottom of it. At some point, I saw a random video of a guy from Paris, who turned out to be a Pandascore employee, explaining why one specific guy is the best in the world in playing video games. I immediately loved the idea of connecting computer science and video games. So thanks to microscopy data deletion, I got a job at Pandascore after my PhD!

What exactly are you doing now and would you say that you found your dream job?

My company is focused on esport competitions – professional competitions in video games. The entire workflow consists of looking at competitions with artificial intelligence (AI), extracting relevant information, creating odds based on these models, and then selling that to clients. My part of the work is developing models with AI, i.e. looking at things such as how many times a specific player died, which strategy each player uses, etc. At the moment, I really do like what I’m doing. However, in an ideal world and in a different social context (higher salary and more respect), my dream job would be teaching because I love conveying knowledge, as well as the thought exchange that follows.

Looking back, is there anything you would change when it comes to your education and research?

It’s tempting to say that I’d tell my younger self to focus more on studying computer science than biology. However, changing one little thing in the past might also change where I am today… and that’s something I wouldn’t want, as I like the place I’m at right now.

Summer School of Science

Knowing your story, it’s not a huge surprise that we managed to bring you to S3 as a project leader three times in a row – at S3 camps in 2017, 2018 and 2019, organised by Dora, Sebastijan and myself. I do remember that you had some doubts before joining us for the first time in 2017 – why was that?

Alex leading the team-building session on the first day of the S3 2019 camp.

I first met Sebastijan at a machine learning summer school in Cadiz. A while later, he asked me to join S3, and that was the end of my summer vacation (just joking!). I immediately said yes to doing a neuroscience project. However, then I heard of the budget limitations, and suddenly I had no idea what I could actually do – I worked in a lab that spent thousands of euros per experiment, using a microscope worth half a million euros, which is nowhere close to the S3 budget. Then a friend from a lab mentioned experiments with bees and I thought it’s a cool idea, though I honestly didn’t know much about bees, so I was doubting whether I could pull it off. I was then trying to give up on leading a project, but you and Sebastijan convinced me to join nevertheless.

We were thrilled to have you in Požega, so thanks for being susceptible to our persuasions! Could you summarize what your three projects were about, and how you found those three experiences?

Alex’s S3 2018 group during the field work with the beekeeper.

The project in 2017 consisted of three parts: building a maze and recording the behaviour of bees, investigating the tongue reflex in bees, and looking into the Stroop effect (the test where you have to choose the colour of the word instead of what the word says). In practice, the project consisted of programming and field work done with a beekeeper, and I had a feeling that the participants preferred the latter. I remember it was an extremely hot and sunny summer, so I bought 6 bottles of sunscreen (because it was so cheap in Požega!) to use during the field work, as well as during the field trip. Also, I expected the participants would be just regular high school students, but they were so smart! I was super impressed how they learned Python in half a day, while it took some of my friends days to reach the same level.

The 2017 experience was so awesome I immediately signed up for the next year, when I led a project with my colleague Eduarda. The project was once again on bees and consisted of the experimental part, data analysis, and modelling the bee brain, so it was really like a research story for a scientific paper. I felt that the students were perhaps a bit less motivated than the year before because there was less field work. The staff that year was the best one ever, especially because of our end-of-the-day (or actually, beginning-of-the-day) meetings.

After S3 2018, I felt like I needed one summer to just chill because I was super tired. However, you managed to convince me to join once again, and my condition was to do a project with Sebastijan. So there I was again at S3 2019, leading a project centred around an EEG headset. That School felt a bit different than the previous ones because the participants in my group, but also in general, were more mature. That year was also the first time I stayed at the Croatian seaside after S3 to get some rest.

Alex’s S3 2019 group setting up the EEG headset on the invited speaker Vladimir Dzyuba.

Do you have any S3 anecdotes you’d like to share?

Helping participants with 3D printing.

There were so many little things that I found amusing during these three years. When I first joined S3 in 2017, it was a bit stressful because I didn’t really know anyone except Sebastijan (and you through Skype), so I was trying to learn and follow the rules. On the first day, we went to Sebastijan’s place to pick up tools to build a maze for the bees, and among them was a drill. So you told me, in a very serious tone, that only I should use it, just to be safe – two hours later all the kids were drilling because we were already behind schedule. In these two hours, I went from I should follow all the rules to there are no rules, this is war.

I also remember how I won the dirtiness competition that Viktoria organised by having (convincingly) the most bacteria on my laptop; when we lost our briefing presentation because the computer rebooted due to a power outage during a storm in 2018; Cedríc running experiments at night like a machine; Elia having 20 pairs of socks that are all the same just to make his life easier; baking croissants at Sebastijan’s place on our day off; how Ivana, the cook, smiled every time she saw me (I was always her favourite because I could eat the most) and her pride when I would ask for a third portion; using frozen peas to put bees to sleep because we couldn’t find ice…

S3 2017 group photo after the final presentation.
S3 2018 group photo after the final presentation.
S3 2019 group photo after the final presentation.

So many awesome moments we had at the camps, all put into a single paragraph… Did you stay in touch with many people afterwards?

It depends on how you define staying in touch. If we say it means that I’d contact people if I were to end up in their cities, then yes, I’d definitely contact the Italian crew, project leaders (especially from 2018 + Elia, Vova, and Anna-Maria), Ana, Vito, Eloise… I also do have a chat where I share scientific memes with Ana and Vito.

Overall, what do you think about S3 and would you like to come again in the future?

I absolutely loved S3! After each camp, I felt really good because I spent time with amazing people and I sort of did “my good action for the day” – it’s not the kind of place where we’re taking money for the organisation and are trying to make a profit; instead we work with little money, but are all passionate about science and teaching. If I had twice the amount of holidays, I would definitely consider coming again as a project leader. This was way easier when I worked in academia because S3 could be counted as working days instead of holidays, but these days I wouldn’t be able to pull that off. I would also love to join as a speaker, especially if I can do it remotely (I’m trying not to take airplanes anymore to protect the environment).

What messages would you send to S3 alumni and to students thinking about applying to S3?

To students thinking about applying, I’d say definitely go for it because there’s not a chance you’ll regret it. Once you’re at the camp, try to live the full experience, e.g. don’t avoid hanging out with your peers, staying up late, playing games, etc. As for the general advice, I’d say never restrain yourself from asking questions – there are people who use tough words to sound smart, so just ask whenever something is unclear.

Thank you for sharing your story with us, Alex! I’m sure many young students and scientists will find it inspiring and encouraging.

Have a question for Alex? Don’t hesitate to write in the comments section or contact him directly – Alex is always ready to discuss anything, from random questions to scientific memes.

By Nikolina Šoštarić

Nikolina is a chemist turned computational biologist, with her current work focused on protein modifications and interactions. Her favourite hobby is painting, and she is passionate about science popularisation.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.