This post is a (likely one-off) departure from the technical themes of previous posts to focus a bit on the career trajectory and prospects of academia compared to industry. I have had the opportunity to develop a lot of very marketable skills in particle physics research. The chance to manipulate and analyze some of the biggest datasets in big data is itself a great resume booster. Combine that with the programming, analytical, and mathematical skills one needs to excel in particle physics, and potential industry jobs are plentiful. Toss in a bit of machine learning, and the ability to distill and convey information effectively, and you’re quite marketable indeed. Top it off with a Ph.D. and a track record of mentoring students, and suddenly you’re a bit of a datascience unicorn.
So why not become a data scientist? The problem, of course, is that having invested so much time and effort in an academic career, why would a a physicist decide to leave academia and move into industry?
Most academics are going to say it’s a bad idea to move into industry, and consider it selling out, because industry lacks intellectual freedom and doesn’t advance humanity’s understanding of fundamental science. Put more bluntly, industry is for profit (i.e. inherently bad) and considered to be boring work. Well, possibly, but it seems like a poor analysis to base my conclusions on an implicitly biased dataset, especially when the bulk of the world works in industry jobs of one sort or another, and perform critical roles in society.
In my experience, if you have the skills to thrive in industry, you end up doing an industry-like job, just in academia, with academic pay. Personally, I typically end end up being a hybrid software developer, systems administrator, and, whenever I can get close to the actual physics data, data scientist. In academia these contributions are appreciated, and personal, but they’re not valued in terms of competitive compensation, and performing them well is unlikely to lead to promotions or greater compensation. This is disappointing, because academic pay is absurdly low compared to industry, and tends to be based purely on seniority, advancing only after time served. After all, no one is in academia for the money; people do this job because they enjoy it. Right? It really comes down to what about academia one personally enjoys, and if it’s the work there are always industry analogues for highly technical experimentalists.
One can also consider the purported perks of academia, like flexible work hours, some latitude in the projects one chooses to work on, and the promise of tenure if you land one of those sweet faculty positions. But the truth is much dimmer when one considers that these perks come with the price of significantly reduced pay, and only a very slim chance of obtaining a tenure track position, which is quite necessary for that lattitude in intellectual pursuits. Further, its likely that several long-distance moves will be required to work several short term positions before having a chance to compete for the stability of tenure, upending your life each time. And now, after a global pandemic upended work expectations, many industry jobs (especially in tech) have transitioned to remote work and flexible hours, muddying the divide between lifestyles.
So, basically, if you’re going to do the tech / data science / machine learning work in academia, you might as well move into industry and get paid for it. The work is similar, the techniques are the same, and the skills are highly valued. If you’re in academia purely for the fundamental research, or rely on someone else to do your data analysis and maintain your computing infrastructure, industry may not be for you, unless you join some well-funded R&D division. In light of this conclusion, would I again pursue an advanced degree in Physics instead of a more direct route to industry/datascience? Absolutely! My time in grad school was invaluable for developing skills and gaining exposure to techniques. That said, I might have skipped a postdoc position and gone directly into datascience, were I to do it all over again.
>> Home