In this guest post, 80,000 Hour’s member Ben Kuhn describes how he looked for a job after graduating from Harvard with a maths major. Ben’s especially reflective, so it’s fascinating to hear how he went about choosing between options in software, startups, finance and research with the aim of making the biggest difference.
For my first few years of college I prioritized getting experience in a bunch of different potential fields–I tried working at Fog Creek, Jane Street, and GiveWell, and cofounding a startup. By the end of that I came to a couple conclusions about what I wanted to do.
In terms of altruistic career choice considerations, I decided I should probably focus on doing the things I could be most awesome at, rather than trying to naively maximize earnings or maximize direct good done–basically, because I’m fairly uncertain about whether having lots of money will be helpful, and I’m fairly uncertain about what does the most direct good, but being awesome at things is a robustly good outcome that can be parlayed into many different advantages later.
Historically, technology- and software-related things seemed to have some of the greatest potential for me to be awesome at them, and also the widest breadth of opportunities to improve the world with those abilities later, so they seemed like the most promising options to pursue further. But I had already done one software internship, and while it was a fun experience, I didn’t want to do anything very similar–I guessed that I’d hit diminishing returns for standard software-engineering internships.
I was concurrently in the process of realizing that studying at Harvard for a fourth year didn’t seem especially high-value, and that I could graduate in three years if I wanted to thanks to my Advanced Placement credits. So I realized that I needed to put a lot of effort into my summer job search to make sure I found something that wasn’t a repeat of my previous internship, and that I would be happy turning into a full-time job if I decided I didn’t want to go back to Harvard.
I looked for opportunities in a bunch of different ways:
I started by listing a bunch of companies that I remembered hearing about that might be especially interesting.
I tried to learn about other interesting firms by following Y Combinator and looking through the portfolios of some venture capitalists that I thought had good taste, e.g. Founders Fund.
One of my roommates had interviewed for an absurd number of internships and probably knows half of the recruiters in Silicon Valley by this point. He gave me a bunch more excellent recommendations, as did some of my other roommates and classmates.
I tried to find companies that were known for being selective or having a difficult interview process, so that I could calibrate my assessment of how good a candidate I was. (In past years I had difficulty with this since I had only completed a few interview processes.)
I had previously made a number of older friends who were well-connected to technology companies through the rationality and effective altruism communities. I wrote to them asking if they had any leads on interesting job opportunities. I assiduously followed the chain of references they gave me out to about three degrees, using Boomerang to make sure I didn’t lose any follow-ups.
At the end of all of these I had a list of about 20-30 companies in a bunch of different fields, which I then had to prune.
The next step was to figure out what I actually wanted from a job. The criteria I eventually decided on were:
Learning–especially since I thought I might skip my fourth year of Harvard, I wanted to be sure I would continue to learn things wherever I ended up working.
Responsibility–I wanted to have a good chance of being responsible for big things, because I’d be better able to make meaningful contributions to my employer’s success. That seems important for accumulating career capital, connections, etc.
Importance–I wanted to work at a company that seemed likely to succeed. I didn’t have very complicated heuristics for “likely to succeed”, but I generally tried to avoid consumer products because consumers are crazy and I don’t understand them. Plus consumer products are more visible, hyped and glamorized, hence over-subscribed.
In defiance of Paul Graham, I also avoided ideas that seemed dumb, and focused on things that sounded like they were solving real problems or fixing real inefficiencies.
I think these were all pretty straightforward criteria. The only notable omission is probably salary: I’m not sure that in the long term I want to use my tech skills to earn to give. And even if I do, my lifetime earnings depend as much on the career capital and skills I develop right now as they do on my salary. So I didn’t think it was that important to have my first job be high-paying.
In the end, after considering dozens of firms, I applied to five, and had a few more on my to-do list when I decided to accept an offer and end my search.
The application processes ranged from extremely rigorous (one finance company flew me to New York for a full day of one-hour grillings on math problems) to fairly casual (one startup gave me just a couple Skype interviews–more understandable since I was applying for an internship originally).
I got offers from all five and ended up picking Theorem, a start-up that uses machine learning to predict loan repayments and give borrowers better prices. This was for a couple reasons:
It was the smallest place I applied to, so I’d have the most responsibility.
Their product was an investment vehicle, which means that they punch above their weight in terms of product importance: they had only a few customers, but my code would be handling millions of dollars’ worth of investments.
Most of my work would be on machine learning, which seemed like a powerful toolset that I wanted to learn more about.
They had a good case that they were especially well-positioned to fix a large market inefficiency, which bodes well for their chances of success down the road.
About a month after I started, I decided that working for Theorem was in fact cooler than doing another year of college, so I sent the cofounders a nice email recounting all the progress I had made for them and suggesting they hire me full-time, which they promptly did.
The quality of opportunities that I found seemed pretty random–for instance, I think Theorem was probably at least twice as good as my next-best option, but I could easily have missed out on finding them entirely if I hadn’t remembered to email someone who told me to check back in three months!
So it probably would have improved my chances of finding something awesome a lot of I had been able to look for opportunities in even more fields–for instance, if I had spent more time during undergrad building related skills, like robotics, electrical engineering or computational biology.
Theorem was one of the firms I found by asking friends of friends–actually, the cofounder of Theorem was three or degrees of separation from me. I suspect that many of the most exciting opportunities are found through connections rather than through ads or Web searching, so it pays to really pump your network during job searches even if you could just rely on Internet searches and random word-of-mouth (as I’d previously been doing).
Of the other place-finding strategies I tried, probably the least effective was listing places I’d heard about before. This might be an artifact of my search criteria, though–I ignored a lot of the best-known internship programs (Google, Facebook, Microsoft, Amazon) because I thought I wouldn’t get to learn very much or be responsible for anything important.
I suspect that asking to convert to full-time after I’d already been working at Theorem for a month weakened my negotiating power somewhat when we were discussing compensation, because I had no competing offers at the time. (On the other hand, I probably got additional leverage because I had done some fairly important work for them already. I’m not totally sure which effect dominated.)
Interviews took up more time and attention than I expected. I’m glad I only interviewed at five places. I wish I had been more aggressive about cutting off interview processes when I didn’t think they would affect my decision, and starting up new processes at other places instead. For instance, I kept interviewing at one New York company after I had basically decided that I wanted to work in the Bay Area, and I don’t think it was worth the time.
I think I could have been more efficient about using the interview process to learn about the places I was looking at. Part of this might be that phone screens are just a lot lower-bandwidth than actually visiting the office. Part of it was also that I didn’t have a good set of things to look for or ask my interviewers.
I’m sure I’m leaving out a lot of the most interesting parts of this process. Do you have other questions about how it worked? That’s what comment threads are for!