I can’t imagine going back
Published:
Spoiler: I could :)
A bit about me: I left the tech industry after 6 years, arguably at the peak of my career, to go back to grad school. When I shared this decision with my friends, after the initial congratulations and “I’m so happy for you”s, I often got a few questions: 1) why do you want to go back? And 2) what are you researching? I assume they were asking out of politeness or to keep the conversation going, but a few years into my Ph.D. I thought I should finally sit down and answer the (arguably) easier one: the first.
If you’re here reading this post, you’re in one of a few camps: you’re contemplating grad school yourself after a few years away working somewhere, to which I say know that my experience is as someone who worked in tech and does computational neuroscience (highly related to my undergrad and what I worked on in industry), so YMMV; or you’re a friend who knew that I waffled through my in-person response and you’re hoping that I’ve fleshed out my thoughts; or I just shared this on LinkedIn or one of the grad school subreddits, in which case hi!
Context
To properly answer why 2022 felt like the right time, we need to go back another 10 years, to 2012, when I first applied to CMU. At the time, a standard college application essay prompt was “Why do you want to join $COLLEGE_NAME?”, and for my CMU application I wrote about wanting to build brains from computers. The exact details I no longer remember, but it was something about researching how we could hook up enough CPUs together to simulate every neuron in the human brain simultaneously. Lofty ambitions for a kid who hadn’t even written his first line of Python, but the admissions office liked it enough to admit me into the Cognitive Science program. Fast forward to 2017, when I graduated, and I felt like I was no closer to accomplishing my original goal. The only thing that had changed was I realized how lofty and underspecified my research question was. I was at a fork in the road: I could cut my losses and go into the tech industry, or I could go to graduate school. As I was mulling taking the GREs and applying to masters programs (my undergraduate GPA was abysmal after floundering around for my first two years), I ended up having a long chat with one of my research mentors at the time, which changed my trajectory through life. He sat me down and very kindly shared his own journey from his undergraduate degree to his Ph.D. His was slightly less meandering than mine, but he worked for a few years after graduating and then applied to graduate school once he was sure that it was what he wanted to do. He then gave me two insights to reflect on: the things I was interested in researching just weren’t in line with what the field was working on, and I wasn’t “ready” yet; even if I was accepted, I would probably drop out of the program. Ouch. Rough, but in hindsight, true.
It’s All About Timing
Claude Shannon was a genius and his seminal work on information theory (which warranted a Ph.D.) seeded an entire new field. I am not Claude Shannon, and there was no way that I would be able to spawn an entirely new field around neuroscience-driven deep learning. So, I was left with a choice: apply to a professor who was working on the topics that interest me and pray that the subfield takes off, or put off my Ph.D. ambitions until I was confident that it was stable enough to sustain itself, but still early enough that I might be able to make a contribution to it. That’s a lot to hedge against, and I don’t think it would have worked out because (as I’ve learned over time). Perhaps more importantly, the questions that I found interesting when I graduated, I no longer find that interesting (or even answer-able); I no longer care about creating an artificial brain and I’ve instead become much more interested in studying structures in the brain and how they influence the kinds of computations we can do. It’s fair to say that my interests have changed since joining my current lab, the Ahmed Lab, but even then all my work is possible only because of pure dumb luck and good timing; as I was gaining faith that my field of research interest wouldn’t disappear overnight and applying to labs, the D. melanogaster community was having its own “upheaval” with the release of the Flywire dataset. The dataset was a marvel of modern ML and intensive neuroscience work: the community worked together to map out the entire connectome of the fruit fly. If it weren’t for the release of this dataset (which opened up far more possibilities than my original reasons for joining the program), my research would look very different and I don’t think I would find it as interesting.
Pure dumb luck.
You Have to be “Ready”
There’s something to be said about timing, but there’s also something to be said about who you are as a person and where you are emotionally. Looking back, I know that I just didn’t have what it would take to make it. Too much of my self-worth was tied to my research output and I just took failures far too hard, which is ironic to say coming from a Ph.D. trainee. What I mean is this: you need a sort of “balance” where you have to care about your research, and I mean deeply care about your research - it has to fill you with wonder and make you really contemplate the hard questions - but at the same time you have to be able to take your failures (and there will be many) on the chin and try the next idea. In college I was a solid B- student (a solid C- student up to my junior year, when I started research), so my sense of purpose was intimately tied to my research and I couldn’t imagine life outside academia because so much of my “upturn” in confidence was tied to research. By taking a step away for a few years, I learned that I could very much exist outside of the research sphere. In fact, I could thrive, and this has honestly carried me through graduate school.
Returning to Academia
Before I begin, I have to state that I unequivocally believe that there are fields and industries where you can get by without a Ph.D. I knew extremely talented people who went on to work as researchers or research engineers at DeepMind or Google Brain after their undergraduate degree or their masters. I consider this the “best case” - you have a large moat of resources, fantastic pay, and access to world-class researchers all working on similar things. And then there are industries where the work doesn’t tie as directly with academia, and here you’re left with the unfortunate decision of whether you should return. I think neuroscience is one of those subfields. When I first joined the lab, I was thinking about how we can use curiosity-based RL methods, or inverse RL methods to model behavior and use that behavior to better understand these neural systems. Over time, my work and research interests evolved into studying the relation between structure and function (see my lab’s webpage); without having access to the lab materials (reagents, gases, etc.) my work would be impossible. My work fundamentally would not be done in an industry setting, not because of a lack of resources, but because this sort of work isn’t yet marketable and there’s no clear path to monetization. Also, a lot of my work is researching some hypothesis about how manipulating some gene will influence behavior, which isn’t in line with how a company works and what a company does.
So, why did I return to graduate school? I returned because the kinds of questions I was interested in, I could not do alone and was very unlikely to do at a company. Why did I choose that year to return? Around the time I applied back in 2022, there were quite a few papers that guided my thinking and reasoning: Reinforcement Learning, Fast and Slow, Liquid Time-constant Networks, and In vitro neurons learn and exhibit sentience when embodied in a simulated game-world. It wasn’t so much that any one of those papers signaled to me “now’s the time”, but more that it seemed like research into what I was interested in weren’t just niche one-off papers, and gave me confidence that interest in these topics wouldn’t just dry up. These papers signaled a resurging interest in bringing biology, neuroscience, and machine learning together; whether to study behavior that’s extremely difficult to capture in the wild, to train more efficient models, or just crazy sci-fi stuff that it felt like the field had lost while chasing benchmarks. From attending NeuroAI Seattle in both 2024 and 2025, I think that my hunch was right - hearing the speakers talk about the interesting work they did made me confident that the field had a strong future.
The Opportunity Cost of Graduate School
Let’s get one thing clear: the opportunity cost of grad school is immense, and it only gets worse the further out you are from your undergrad. From a career progression standpoint, the worst part is that the post-Ph.D. pipeline doesn’t look much better - excellent researchers struggle to find positions all the time. From an earnings standpoint, simple back-of-the-envelope math puts the loss of earnings at over a million dollars in raw salary, not accounting for promotions, benefits, and compound interest. I like to think that I live rather simply, but even then, it’s still depressing how slowly my bank account numbers go up. Which segues us into lifestyle changes: you’re taking on (far) more work and stress, and will likely strain many of your friendships (hopefully all that time you spent in the years before set you up with a strong group of friends!).
From a social standpoint, you will undoubtedly experience FOMO - your friends will be getting married, buying houses (fingers crossed) and hitting all the “milestones” of being an adult, while you’re “still in school”. I understand that academia has a reputation for sticking to itself, but I get it - the cadence your life follows is just fundamentally different from that of your friends. To people with no insight into the lifestyle, you’ve entered this phase of “delayed adulthood” (a comment I actually received), but in reality you’re just trading off certain things for other things. This might be controversial, but having worked in the software industry, I can and will unequivocally say that working in academia is far more stressful. At the Ph.D. trainee level, every moment you’re not researching is time that another researcher could be spending working on a similar idea (which might lead you to not getting that post-doc position) and at the PI level, you’re constantly worrying about funding for your lab and endlessly writing grants.
Closing Words
Despite all these drawbacks, I will say that I think everyone should work for a few years outside academia before going back. Having worked for a bit undoubtedly changed my perspective vis-à-vis time management, productivity and self worth. If you’re working and have left academia for a bit, I’d say reflect on what I mentioned, particularly about doing research while working at a company. If you truly feel that you cannot accomplish it in that context, then I’d say take the leap. If you’re a friend who is reading this, do let me know what you think and hopefully this answered any questions you might have had!