This is Bob.
We’re hiring
It’s that time of year again and we’re hiring at all levels at the Center for Computational Mathematics (CCM) at Flatiron Institute (the in-house research arm of Simons Foundation). As they are listed, job ads will appear here:
Our mission
I finally work at a place whose mission aligns with what I actually do.
The mission of the Flatiron Institute is to advance scientific research through computational methods, including data analysis, theory, modeling and simulation.
This place is entirely researcher driven. This is the only top-down directive I have, and I can choose what I work on as long as I stay productive and on mission.
About Flatiron Institute
In addition to CCM, Flatiron has centers devoted to computational biology, computational neuroscience, computational astrophysics, and computational quantum physics (not to be confused with quantum computing). It’s a very problem-rich environment, because all these domain scientists have lots of interesting data (for example, I spent an hour yesterday talking to a CCB researcher about his data involving flies jumping when exposed to light, and we went over how to build hierarchical Bayesian time series model and I wrote some Stan code to match). Each center is comprised of roughly 60 people, around 30 of whom are full-time research scientists, 20 of whom are postdocs, 5 of whom are software engineers, and 3 of whom are dedicated admin staff.
We also have a dedicated high-performance computing group, the Scientific Compute Core (SCC), and a massive (for academia) compute cluster with hundreds of H100 GPUs and hundreds of thousands of CPU cores.
It’s all on site
We do not hire remotely for full-time jobs. Luckily, our offices are really nice, bright, and well maintained from IT to physical facilities.
P.S. “Flatiron” is the name of the neighborhood—our offices are two blocks south of the Flatiron Building, in one of the nicest areas of Manhattan (a short walk to NYU, Google, and Meta).
About CCM
Our work can broadly be categorized into traditional numerical analysis and scientific computing, computational statistics, and machine learning. In all of these areas, we work on theory, algorithms, and release open-source software. All of our research and external funding is funded through the $5 billion endowment for Simons Foundation—we are not allowed to apply for grants. As an aside, Simons Foundation hands out over $300 million in science funding per year.
We have really great software engineers; in particular, Brian Ward and Steve Bronder work with me on Stan-related projects and with other people on other projects across Flatiron. There’s much more of a career track here for software than in academia, and it’s allowed us to recruit really top-notch people.
We also have a fantastic admin staff, which makes an incredible difference.
Flatiron research fellows (aka postdocs)
We hire postdocs for three years. We’ll be trying to hire around 6 postdocs this year across the areas of CCM interest. I’ll be out of postdocs after this year—Yuling Yao started a faculty job at UT-Austin, Charles Margossian is on the job market, and Sifan Liu (technically visiting faculty) will be starting her faculty job at Duke next year.
In general, we realize that our postdocs can’t strictly follow our mission if they want to get an academic job, because academia still doesn’t value software contributions very much. Nevertheless, I would very much like to recruit someone who’s just as interested in contributing to open-source software as publishing NeurIPS papers. One of our recent postdocs, Wenda Zhou, who did his Ph.D. in stats at Columbia, used his time here to retool as a neural network researcher and ML engineer—he now works at OpenAI.
Joint faculty position with Cooper Union
This year, we’re looking to hire someone for a joint position with Cooper Union, an engineering-focused college in NYC. Here’s the job posting:
- Cooper Union Computer Science /Flatiron Institute CCM: joint tenure-track faculty position
It’s 3 years of a 50-50 position, then transitions to “full” time at Cooper Union (scare quotes because academic jobs in the U.S. are typically 9 month jobs). That’ll be good for tenure decisions because academic departments have a hard time with split commitments.
Research scientists
We have a few openings across ML and statistics for permanent research scientist positions. We’re looking for people with at least a few years experience post-Ph.D., but will consider people just finishing a first postdoc. We can also hire much more senior research scientists like me or Lawrence Saul, our head of ML. I would ideally like to hire someone who has worked on both computational Bayesian statistics and on neural networks.
We’re still tweaking the job ad, but it’ll go out any day now and show up on the list posted above.
Let me add that, yes, Flatiron is a wonderful working environment. I go down there often to talk with Bob and his colleagues, and lots of great ideas have come out of that.
Is there a reason why you choose to ignore the latest 20 years of advances in technology (not to mention the past 4 years of empirical experience) and forbid remote work? An onsite-only policy excludes lots of disabled and neurodivergent people (not to mention those who can’t stomach Manhattan rents/commutes). But surely that wasn’t your goal?
Remoter:
I can’t speak for Flatiron but, in my experience, including in the past 4 years, it really does make a difference to work in person. I’ve found in-person work to be important both for teaching and research.
That said, I’m sure that there are people who work better remotely. And, as you say, lots of people don’t want to live or work in NYC. Fortunately, it’s a big country and there are lots of jobs outside of the city. It makes sense that the Flatiron Institute in downtown Manhattan will attract people who want to work in this location, and it makes sense that people who’d rather work elsewhere or work remotely would look for jobs in other places.
I can’t speak to the Flatiron Institute’s experience, but research shows that 100% remote work is associated with lower job satisfaction.
Among the conclusions from a survey of 10,000 people, “Those who work in-person or in hybrid work environments have higher vitality and are more satisfied with their jobs than those who are fully remote. Hybrid workers also report healthier habits than remote workers, which results in better mental and physical health, greater levels of exercise, and at least one hour a day spent outdoors.” (https://newsroom.cigna.com/retaining-high-vitality-workers)
In that link, a lower percentage of remote workers said they had job stress than either hybrid or in-office.
A 12,455 person study found a 20% higher happiness among remote workers: https://www.forbes.com/sites/bryanrobinson/2022/05/05/remote-work-increases-employee-happiness-by-20-new-study-finds/
An academic study (smaller N, but experimental design) found higher happiness for remote workers: https://pmc.ncbi.nlm.nih.gov/articles/PMC9368397/
A large meta-analysis found that remote work was associated with greater job satisfaction: https://onlinelibrary.wiley.com/doi/abs/10.1111/peps.12641
As Andrew says, people have different styles and not every company or institute has to have the same policies, at least in New York, where we’re located.
Absolutely everything’s accessible here and Simons Foundation goes out of its way to hire a neurodivergent staff (and I’m not just talking about the physicists and mathematicians)—more so than I see elsewhere.
Perhaps not surprisingly, I was not involved in the decision about everyone having to be on site. The goal, in my understanding, is to make the place more pleasant as a genuinely interactive working environment. I get lonely after too much time remote—our offices close for 2 weeks in August, and it’s long enough that I miss going into the office.
When I worked at Lucent Bell Labs in New Jersey and nobody came in, it was a terrible environment. The stats department at Columbia’s a ghost town now, like a lot of academia. When people stop coming in, it creates a cycle where eventually nobody comes in, because there’s nothing to come in for. UMass Amherst CS has gone so far as to allow their full-time faculty to live somewhere else and teach their classes remotely. I’m curious what universities are going to do with all their wasted space now.
We lost some good people after the pandemic due to the decision to go fully on site. I believe we also recruited some people due to the vibrant community here on site every day.
My own productivity is so variable day to day and week to week, and the outputs are so intangible (research code or papers), I don’t even know how I would measure them. Keystrokes? Lines of code? Pages of notebook filled? Actual good ideas? I think were you to measure peak idea generation, the shower or half awake in bed would be the most productive place I spend my time! I find my perception of my productivity and my enjoyment (strongly correlated) depends on what I’m doing. If I’m doing math or algorithm design with someone, I strongly prefer sharing a board in person. If I’m coding with someone else, you can’t beat sitting down together and pair programming. If I just have bricklaying type API coding work or deep thinking about code or math to do, I prefer to do that at home after 11 pm when there are zero distractions. We can work from home one day/week, but most people choose not to—I only do it when I need five or six hours of uninterrupted concentration. I can’t think of anything I’d rather do online rather than in-person or by myself. Let me say again that these are my own personal preferences. If I worked somewhere with a lot of online staff, I don’t think I’d be happy—productivity isn’t everything :-).
I also play a lot of tabletop RPGs both online through Roll20 and in person (as in multiple times per week). Online’s the only way I get to connect with my childhood friends, but playing in person is soooo much more fun.
I’m happy to vouch for Flatiron. It got on my radar when I noticed people whose work I follow kept turning up there. I visited for 10 days in September. The researchers, leaders, staff and visitors are amazing. You get to talk about really interesting problems with smart people, without having to break to go to a class or a committee meeting. There’s a great mix of skills and interests, so you’ll learn new things. I had numerous long technical conversations; sometimes they ran so long that they caused late lunch or missing a cool seminar or even another good technical conversation, but you just can’t be two places at once. It seems like there is only the bare minimum of overhead tasks to do. It is also in a very nice part of NYC.