MSR

Many of my friends accept lower salaries for work benefits such as location preference and work-from-home flexibility. How much of a paycut would people accept to have intelligent and inspiring coworkers? I've been mulling over this question all summer.

I think what impresses me the most about MSR's DSG is management's technical ability. Their questions and suggestions leave no doubt that if I were hit by a bus tomorrow, they would be able to pick up where I left off. This is in stark contrast to anecdata from my industry friends, many of whom labor under professional non-technical management. With technical management, I feel sharper after every meeting: more precise, more concise, a better grasp of the big picture. For example, my favorite feedback from this summer was after I gave a long presentation: "I notice that you are very carefully not telling me about (obscure aspect X), that's fine, but can you tell me why?". To me, it seems like the difference is this. Non-technical management can help you decide between options A, B, and C. But technical management will think of option D (and ask you why you're not doing that instead). To grow, especially as a researcher, technical management is a blessing. I actually still find it difficult to think of my mentors as managers. I first learned of their names as "those people at MSR that keep publishing solid work in the ML for DB space".

More broadly speaking, a significant perk of MSR is that there is a culture of good science. There were opportunities this summer to boost performance at the cost of scientific rigor, and the message at all levels was clear: pick scientific rigor, set that dodgy feature aside for now and assume that we won't have it. Beyond my own experiences, I met multiple new full-time hires across MSR. A common theme in their starting instructions: take a few months to talk to people across the company and figure out what is the right problem to solve. Where will you find better conditions for a medium-term high-impact bet? A high-trust culture like this is rare. Importantly, the researchers there earn and deliver on that trust.

On the topic of trust, management solicits and acts on constructive feedback. The DSG leader asked what the interns wanted and was quick to deliver. More face time with the full-time researchers? We had lunch sessions organized by the next week. Someone asked for healthy snacks? Apples and oranges appeared on the common desk (though heaven knows we could have bought our own), along with some candy bars. Of course, not all asks can be granted (nor do they make sense to grant, in my own opinion). For example, cross-organization asks are not entirely within DSG's control, and anything that consumes group-wide attention requires more justification. To me, the key aspect is that the reasoning behind each outcome was clearly explained.

At lunch, a long-time employee shared an interesting perspective on what distinguished MS from other companies that seem to have lost their focus (couGh). That is, the DNA of MS is technical: look at the managers, and you'll see someone who earned the respect of their peers as an individual contributor to get there. While many companies pay lip service to this idea, it seems to be actually true for MS. Even their "non-technical" CEO scored higher on the Putnam than their technical founder, which is wild.

Without other constraints, I would seriously consider one of these two options. First, disappear from tech, start a capybara cafe, and just retire from life. Second, go right back to DSG under this leadership, develop myself as a researcher, and aim to become like the people I admire in 20-30 years.

But there is another calling in life, where I feel like if I do not do it, then nobody will. We will see.