Recurse Center

🎬 Tomo Bruce Hill recently gave a talk at Recurse Center showcasing Tomo, a programming language he’s been building as a personal project. If Bruce’s name sounds familiar, it might be because he also recently helped bring the Stainless CLI generator to GA 🎉. It turns out many of the ideas he’s excited about show up clearly in Tomo as well: reducing incidental complexity and letting tools do more of the work for you. Some highlights from the talk: * No separate build system: The compiler handles package management, lazy recompilation, and dependency resolution. Run the compiler, and if you're missing a dependency, it just asks if you want to install it. No separate package manager commands needed. * Integer overflow safety: Default integers automatically overflow into big ints (like Python), but in a statically typed language that compiles to C. Small integers stay in registers without heap allocation. * Built-in CLI argument parsing: Define your main function with typed parameters and defaults, and you get help messages, flag parsing, and even enum support automatically. No third-party library needed. * C interoperability: Tomo transpiles to C, making it easy to tap into the existing C ecosystem. Bruce wrapped terminal graphics, HTTP (via libcurl), and more. The demo? A 150-line Snake game running in the terminal that was built in a single evening! As Bruce put it: "Languages can be more or less than the sum of their parts depending on how those features interact with each other." Learn more: https://lnkd.in/g_nH8b89 https://lnkd.in/gv4U5DDa

Why do people come to the Recurse Center? We’ve been digging into this question recently. My colleague Emily Bernier talked with alums and clustered the themes that kept coming up; much was expected, but some of it was surprisingly raw. Alums: what resonates? What’s missing? Community - A uniquely collaborative, intellectually stimulating environment - Low-ego, non-condescending peers - Feeling socially fulfilled - Feedback and ideas from people working on genuinely interesting things - Motivation from being surrounded by other curious, like-minded learners “Working solo, you don’t know what you don’t know” Structure - A bounded, high-focus period to make real progress - Time you can deliberately defend from other responsibilities - Enough structure to avoid spiraling; not so much that it’s constraining - Space to deeply focus on a single, ambitious project - “I don’t trust myself with that much free time” Life transitions - Between jobs (laid off, job-searching, internship break, sabbatical) - A reset point during moments of change Third space - A physical place to work - Not being on this journey alone at home Job goals - Learning AI - Pivoting (or returning) to tech - Building a concrete, impressive body of work - Becoming fluent and confident in day-to-day workflows - Optional job-search support Legitimacy - “I needed to justify this to my parents” - Working on projects "without having to say you're unemployed" Joy - Building for joy instead of for work - Reconnecting with creativity in programming - “I’d missed the creative satisfaction of coding” Serendipity / something new - “I was in a rut” - Not knowing exactly what you’ll discover - Setting aside career anxiety and recharging - Excitement about whatever happens when a group of curious people spend time together

What's the most fascinating thing you've learned in the past month?

If you're not working or you're planning to quit your job, we're hosting a panel that's just for you: How To Make Your Time Off Count. The event will feature three Recurse Center alums who will share their experience taking time off of work. Funmi Oludaiye is a triathlete and alum who came to RC as part of a year-long sabbatical after a 15-year run at Goldman Sachs. Gage Krause attended RC after leaving a philosophy PhD, as a transition into professional programming at Coverbase, an SF-based AI startup. Frank Chiarulli Jr. came to RC after many years working in startups and finance. Come hear how they used their time off to explore new domains, rediscover their joys, and achieve their goals. https://luma.com/zfhxstgy

Failing to raise venture funding after Y Combinator was the best thing that happened to us. We were one of the only companies in the Summer 2010 YC batch that failed to raise a proper round. After Demo Day, we biked up and down Sand Hill Road talking to many of the top VCs in the world. About half said no; the other half simply ghosted us. We felt like total failures. Nine months later, we managed to raise $270k from a few wonderful angels. It was a relief, but it also felt like a paltry sum compared to our batchmates, most of whom had raised millions. At the time, we thought we were building a venture-backable software company. We spent a year pivoting and exploring ideas, but none of them really went anywhere. Then, in the summer of 2011, we ran an experiment: the first batch of what became the Recurse Center. People loved it, so we ran another. Over time, two things became clear. First, the company we were actually running looked nothing like the one we’d originally set out to build. Second, the company we wanted to build was human-scale, not venture-scale. Because we’d raised so little money, it was relatively easy to pay our investors back, which we did. Today, the Recurse Center is profitable, independent, and controlled entirely by the people who work there. (YC still owns a small slice of our cap table, and we’re glad they do; they’ve been incredibly supportive over the years.)

"How did you spend your time off of work?" If you're lucky enough to have a chunk of time when you don't need a job, you want to make it count. Most of us get few if any points in our lives when we can do this.   Of course, only you can answer this question for yourself. But it's helpful to hear stories of how others have used their time, and what they wish they'd known when they were in your shoes. Over the past 15 years, we've had thousands of people spend time at the Recurse Center as part of their time off of work. We've gotten to see what works and what people regret. Next Thursday, we're hosting an online panel with three of those people. They will each share their stories of how they chose to spend their time not working. How to Make Your Time Off Count: https://luma.com/zfhxstgy

Many people don’t realize that when you leave a job, you lose more than a salary and benefits. Some of that loss is liberating. The boss you couldn’t stand? Gone. The looming deadline? Doesn't matter. But for many of us, work is also a major source of structure, camaraderie, and purpose. (You can debate whether that’s good or healthy, but it’s undeniably true.) When you leave a job, all of that disappears. The place you went every day, the people you saw there, and the shared thing you were working toward. This is true whether you were fired, laid off, or chose to leave. If you don’t need another job right away, not working can feel incredibly freeing. You have total control over your time! No meetings, no calendar, no OKRs. Wouldn't it be nice if you could keep the good parts of a job? You can. Many programmers have told us that the Recurse Center recreates some of the best parts of a good job, without any of the bad ones: curious, kind, and motivated peers to learn and work alongside; a beautiful and energizing space to go every day; and a shared context in which to pursue your own goals. Even some of the small, intangible things RC provides, like a rhythm for your week and a proper start and end to your "work" day, can have a big impact on your mindset and how much you get from your time off of work.

Any Recurse Center alums have advice?

AI’s impact on programming jobs so far is clear: it's made tutorial-level understanding almost worthless, and deep expertise more valuable than ever. Not that long ago, you could make real money building WordPress sites or Rails apps even if you only sort of understood what you were doing (no shade here; everyone’s got to start somewhere!). You might unknowingly ship 5 MB thumbnails or an XSS vulnerability, but you could still find work, even if you routinely ran up against the limits of your understanding. Now, Claude Code can build generic, tutorial-grade sites that kind of work in minutes. This is one of the few things both AI skeptics and AI boosters agree on. And we’re seeing this play out in the job market: expectations are up across the board and surface-level programming skills won't get you offers these days. At the same time, skeptics and proponents can also agree that a much smaller set of jobs now pays more than ever. At the extreme are the hard-to-believe but very real $20–100M offers for top AI researchers that made the news this year. But there are orders of magnitude more jobs one or two levels below that which still pay exceptionally well. If I had to bet, I’d say there are more programmers making $250k–$1M in 2025 than ever before. There’s no evidence that AI is hurting those jobs, and plenty of evidence that AI (along with its hype, revenue growth, and VC funding) is creating more of them. This isn’t a prediction. I’ve long since given up on those. It’s simply an observation based on what we at the Recurse Center have seen so far. 

Do staff engineers make $130k or $6M a year? The answer is both. There’s 10x, maybe even 50x, more variance in the job market than most people realize. If you’re a job seeker, you usually only have one real data point: your own experience. Anything wildly different from that can feel hard to believe. But when you work with hundreds of job seekers (we’ve helped 350+ this year at the Recurse Center), you see just how extreme the spread really is. In 2025, we’ve seen: - candidates negotiate their offers up by $700k+ - companies refuse to budge even $5k - senior engineers struggle to get offers - college dropouts get multiple offers - ICs making $1M, $2M, even $6M+ a year in liquid comp - staff engineers making $90k/year - remote jobs with 4-day workweeks - in-person jobs with 7-day workweeks - job searches that took three days - job searches that took over a year - a candidate who got an offer after a single phone call - a candidate who got rejected after 17 (!!) rounds As with everything, context matters. The candidates getting eye-popping numbers have deep, specialized ML experience (and some haven’t had a night off in months). The staff engineers making 1/10th as much are choosing mission, flexibility, and work-life balance (or they’re in Europe). Many of the folks struggling to get offers are playing on extra-hard mode: they’re great engineers, but life and family constraints narrow their search. The programming job market isn't monolithic; it’s fragmented. And advice that’s true in one slice of it can be completely wrong in another.