I first met Rani when he came to Imperial to give a talk that from the abstract, seemed to suggest it was about reinforcement learning (an area of machine learning). Like any physicist trying to pretend they belong in a computer science department, I tend to go to every lecture I feel I might have a chance at understanding.
I was pleasantly surprised when I realised, 10 minutes into the talk, that Rani is actually talking about how ideas from RL can be applied to experimental psychology, trying to understand how humans actually think and make decisions.
I won’t further discuss his research here, he does a much better job of that elsewhere. Rani was kind enough, after his talk, to agree to meet with me and discuss how he ended up in his job and in his field, after starting out from a pure mathematics degree.
Our interview
We met a week after his talk, at Rani’s office at UCL. Initially the idea was to go to a coffee shop and sit down, but the afternoon London crowd convinced us that a patch of grass outside would play a friendlier host.
(Note: What is written below is extracted from a transcript of my interview notes. It does not contain quotes verbatim.)
Summary
Like most researchers that I have met during my stints at CERN, Cambridge, and Imperial, my impression of Rani is that of overriding friendly curiosity combined with an impatience that forces you to get straight to the point.
In our 40 minutes together, I found that Rani went from teaching mathematics, to learning how to code, to designing algorithms in the financial sector. Throughout this time his “hobby” was to listen to psychology lectures, a hobby that eventually turned into a PhD, and finally now, a career. When asked why he finally made the jump, his answer highlights what a lot of people, including me feel — he wanted something to be really passionate about. A cliché maybe. But clichés are clichés for a reason.
When I asked him if he would ever go back to industry he said only if he “fails in his current trajectory”.
Below is the full transcript of our interview.
Full transcript
ALI: Let’s start with a little bit of a background, what is your position today at UCL?
RANI: I am currently a post-doc and I have been for the past 2 years right after I finished my PhD at Tel-Aviv University.
ALI: Ok let’s go back to when it began. You started out studying pure maths as an undergrad, then you went into algorithm design? How did that happen?
RANI: Actually I didn’t do that immediately after my degree. I first taught for a few years. I taught maths. It was during my time teaching that I got interested in algorithms and coding. Actually coming from a pure maths background it was not an easy transition you know. We were studying things like “set theory” and then I was there trying to code a new thing here and there. Not very easy.
ALI: To be honest, I don’t really know what an algorithm designer does. What was your job like? how did you get into it?
RANI: I worked mostly for small companies, and when you work for small companies, you do more. So I did more than algorithm design. Basically what I would do is, a company would have a vague problem they wanted to solve, for example in solar panel optimisation or something in high frequency trading, then I would sit down and try to think of a good algorithm to solve it. Once I had that, I would then sit down with the development team and try to teach them how to do it. Sometimes I would do the coding myself. It was very entertaining and engaging work. I liked it quite a bit.
ALI: If you liked it then why swap it psychology? It seems like quite a leap.
RANI: So it was not such a decisive progression in reality. I had always been interested in psychology. During my time in industry I would occasionally take various different Open university courses in psychology. Once class here, another there. Then after a few years they called me and said something like “Hey if you take a few more courses then you can get a degree”. So I did. And then I had my degree. This was all while I was working.
ALI: That’s pretty cool actually. It sounds like it almost snuck up on you. But it’s one thing to take a few Open University courses, and another to enroll in a PhD programme. How did that happen?
RANI: Well after I had my degree I was looking around and found this professor that seemed to be open to what I was interested in. So then I started a part-time PhD. I still had a leg in the real-world while doing my research.
(Note: To all the academics out there, I love how even Rani, himself a current academic, regards “research” as not the “real-world”.)
ALI: Were you looking to leave your work in algorithm design? Why?
RANI: So in all the companies that I had worked, I would notice that some of my colleagues, when they were working, they were really into it. They were very interested in the work and really cared about the work. I also enjoyed the work but not to that level. It was like that when I was studying maths but not in this job. So I was looking for something that I could be dedicated to like that.
ALI: It’s funny that you had this feeling before your PhD and I am testing the waters almost at the end of mine. I’m curious, what was the plan after you got the PhD?
RANI: Well actually, before I completed my PhD, my advisors came into a bit of money and suggested I swap to full time. At around the same time, the company I was working for was going under so it seemed like a good opportunity. Also, I was struggling with my research a little bit at the time, and I thought this would be a good way to give it a final push.
ALI: It sounds like it all came together. A lot of PhD students opt for industry after their studies. What is and was your plan?
RANI: Once I then had the PhD, I enjoyed academic work so I applied to a post-doc here in UCL. The plan is to stay in academia, maybe in the UK if possible but in academia.
ALI: I guess this brings us to now. What would you say is the most enjoyable aspect of what you do today?
RANI: Hmm, I really enjoy the fact that I can think of some idea one day and then I am free to spend time thinking about an experiment to validate the idea. Designing the experiment is not easy and most of the time it seems impossible. I like the search. I like the search and I like the idea generation itself.
ALI: It seems like you had some aspect of this in the algorithm design work no?
RANI: I did yes but with less freedom. Sometimes there was freedom but usually there were some limitations and I think I’m the kind of guy that doesn’t like that.
ALI: So do you ever think about going back to industrial work?
RANI: If I fail in my current trajectory then I will think about it.