From Brains to Bitcoin: Why One Psychology Graduate Left Academia for Industry

Edited by Maedbh King
Darius Parvin, PhD is a recent graduate of the psychology program at UC Berkeley. He studied in the Cognition and Action laboratory under the supervision of Prof. Richard Ivry.

Darius Parvin recently graduated from the Psychology PhD program where he worked under the supervision of Prof. Rich Ivry
Darius Parvin recently graduated from the Psychology PhD program where he worked under the supervision of Prof. Rich Ivry

A couple of weeks ago I finished my PhD in Cognitive Neuroscience, and I’m currently working as a software engineer at Bolt Labs, a cryptocurrency startup. Here, I’m going to give a brief summary of why I made the transition, how it happened, and things I wish I had known earlier!

Going into my PhD, I was definitely open to the idea of staying in academia and becoming a professor. I like the idea of being a scientist and I am fascinated by how the brain works. The prospect of discovering something new as well as contributing to the knowledge of humankind is deeply meaningful and exciting. A lot of people recommended Rich Ivry as a great supervisor and I was interested in his work on motor learning, so it was a perfect fit.

So why am I changing fields? Completely unrelated to neuroscience, in mid 2017, I saw a friend’s online post about how cryptocurrencies are the future. Soon after, I fell down the rabbit hole. What excites me most about cryptocurrencies is their potential to profoundly transform global economics, reducing the overall room for corruption and increasing financial inclusion. Having learned about the 2008 financial crisis and the influence of money on politics, the idea of an open cryptocurrency like Bitcoin being the global reserve currency became the new and most exciting idea for me.

I figured my most transferrable skills would be as a data scientist, so I started aiming for that career, with the goal of getting a job in a cryptocurrency company. In the summer of 2018 I did a bootcamp with the Berkeley Graduate Data Science Organisation. We were assigned to groups of 4 and worked with a mentor from industry on a 3-week project. For some (including me), it was our first exposure to data science and a chance to learn about it. For others, it was an opportunity to produce a polished project that could be included as part of a portfolio. It helped me get a sense for what it would be like to go into industry. I would recommend it for anyone interested!

Another bootcamp I had heard about was called Insight Data Science. Their program is longer (8 weeks) and they only accept candidates who have finished or are close to finishing their PhD. Coincidentally, by the time I was applying, they launched a new blockchain program called Decentralized Consensus. I applied for both programs but only got into the latter. If I had been accepted into the data science program, I probably would have taken it because it seemed like the safer career path given my background. However, in the end, it all worked out for the best because I ended up getting my dream job through that program!

I started the Insight program in March 2019 and put the PhD on hold for a couple of months. The group was comprised of approximately 20 people from diverse backgrounds, but consisted mainly of software engineers and academics. For the first 3-4 weeks, we worked on an independent project on any topic of our choosing. After that, we had professional development workshops to practice interviewing skills and to work on our resumes/Linkedin profiles. During that time, companies who were interested in hiring would come and present. We would then get a chance to demo our projects for the companies we liked most, with the hope of getting a callback for an interview.

I went back to the lab to finish up my thesis, and then around July I got a call from the CEO of Bolt Labs to say they wanted to give me an offer! There was some delay as I had to wait for my OPT visa to come through, but I signed the offer and will start in early September as a software engineer.

General advice for transitioning to data science

Networking/experience

Do internships One of my regrets was not doing a summer internship during the PhD. In hindsight it probably would have helped a lot when I was applying for data science jobs.

Bootcamps The two bootcamps I did were both free. GDSO is organized by Berkeley, and Insight gets their funding from companies who end up hiring people from the program. Aside from the companies they introduce you to, the main benefit of doing programs like these is the opportunity to make friends with people who have the same interests and goals as you.

Ask for referrals Don’t be shy about asking someone to help refer you for an interview. Often, companies give a referral bonus to employees who refer someone who ends up getting hired, so it’s a win-win. Even if the person doesn’t know you well, you can even get an interview from someone who’s willing to say that they previously encountered you and that you seemed like a decent person. I had not received a job offer, I’m almost certain that it would have happened through a referral rather than a cold-call.

Skills

Teach Stats Teaching Psych102 was one of the most worthwhile things I did during my PhD. The only regret I had was not doing it earlier. It was a bit daunting, but it made me a lot more confidence in stats and R.

Switch over to python I used Matlab for 7 years and in hindsight it was a wasted opportunity not to learn python. At first it can be a bit intimidating or confusing if you’re used to Matlab, but it’ll be worth it. Ideally you should have someone around who you can ask for help when you get started at the very beginning.

Classes

VS265: Neural Computation This was my favourite class at Berkeley. Prof. Olshausen gives great overview of modelling neural activity using machine learning algorithms. I would strongly recommend this class if you’re interested in machine learning and neuroscience.

Data 100: Principles and Techniques of Data Science If you take this class, you would be in a comfortable position to start doing most of your analyses in python (and would probably do it very efficiently)!