Blog · November 10, 2026 Guide

How to find cracked engineers, starting with a foosball robot

BL
Bolun Li · 7 min read

You find cracked engineers where the work on GitHub concentrates: the busy projects other strong engineers build on, review and star. Then you read what each person shipped there.

Julius Marminge
@juliusmarminge
May
Jul
Aug
Sep
Oct
Nov

Founding Engineer at Ping.gg shaping async state tooling

Works on async state management tooling, after AI engineering at Nustom. Contributes to TanStack/query, a state management library for the web, and typehero/typehero, a community platform for TypeScript developers.

As of Nov 2026

To find cracked engineers, start where the work on GitHub concentrates. Look for the projects that other strong engineers build on, review and star, and then find the people at the center of them. Read what each person shipped there, and then look at their title, which often matches the work.

The clearest example we know of starts with a foosball table.

Xander built a robot that took the fun out of foosball

Xander Naumenko built a foosball table that plays itself. He put a robot on one side, and the robot can defend, pass and take shots against a person on the other side. In the README for the project, foosbar, he wrote the best line on his GitHub: the robot "has successfully mechanized all my joy out of the game, leaving me free to pursue more important and deeply fulfilling things with my life."

He linked those last words to something even stranger. The link goes to computerraria, a working RISC-V computer that he built out of the wiring inside the video game Terraria. The computer runs at about 5kHz with 96kb of RAM, and it plays a version of Pong that he wrote in Rust. When the game's wiring was too slow, he wrote a mod that reimplements it. We wrote up the whole story in Xander Naumenko built a RISC-V computer inside Terraria.

His day job is quant trading at Jane Street, and you can read that on his LinkedIn. The foosball robot and the computer inside Terraria are only on his GitHub, so we read GitHub first.

What you miss with the usual filters

If you run the normal search for Xander, you get a quant at a trading firm. CPU design and robotics don't show up anywhere in that result.

The other shortcuts miss people too. Followers mostly track how much someone posts, and some people farm stars. LeetCode measures how fast you solve puzzles, and that's a different skill from shipping software that people use. If you filter for big-name employers, you drop everyone at small companies, everyone working for themselves, and everyone at places you haven't heard of. A commit count doesn't tell you much either, so we look at which changes landed and who built on them.

What you see when you read the work

When we read the repositories instead, we keep seeing the same patterns.

Liam Brummitt wrote matter-js, an original physics engine for the browser. In the README he calls it "an original JavaScript physics implementation (not a port)". He created the repository in February 2014, and in September 2026 he merged a fix named fix-compound-inertia, so he's still fixing his own engine twelve years later while other developers build on it. He also built tools for creating and debugging physics worlds in the engine, so he did the engine and then the tooling on top. There's more in Liam Brummitt builds matter-js, a physics engine for the web.

Scott Jehl ended his best work with a goodbye. In 2011 he wrote Respond, a tiny script that made media queries work in Internet Explorer 6 to 8. In 2012 he wrote picturefill, so sites could use responsive images years before browsers supported them. Today the picturefill README tells readers that the project is "no longer needed or recommended. Thanks everyone!" He built a tool, people used it widely, and then he retired it once the browsers caught up. See Scott Jehl built the polyfills behind responsive images.

The third pattern is getting work merged into hard codebases, where a maintainer reviews every change before it lands. You can see it in two engineers on Vamo's Discover pages. Julius Marminge, a founding engineer based in Sweden, is the top contributor to t3code, a tool that lets you control coding agents like Claude Code and Codex from any computer or phone. nate nowack, a senior engineer at Prefect in Chicago, is the second most active contributor to fastmcp, the Python framework for building MCP servers, after its creator.

On Julius's card at the top of this post, look for t3code in his repositories, and then read his garden to see how often he ships.

nate nowack
@zzstoatzz
May
Jul
Aug
Sep
Oct
Nov

Applied AI engineer building MCP framework tooling in Python

Works on Python workflow and MCP tooling at Prefect. Pushes to PrefectHQ/fastmcp, a framework for building MCP servers and clients, and contributes to pydantic/pydantic-ai, a library for building AI models.

As of Nov 2026

On nate's card, look for fastmcp in his repositories, next to his work on Prefect.

Everything we just described is public, and we didn't need a resume to find any of it. We keep more profiles like these at /cracked.

How to find the next one

Start from the work. Describe it in a short phrase, like "Rust type checkers" or "browser physics engines", and then rank people by how closely their projects match it. You can add titles later.

That search turns up people you've never heard of, so the next question is which of them matter to your team. We usually look at your own stack for that. People who contribute to the libraries your product runs on already know your problems, so pull the contributors of two or three of those projects and then filter them by place.

Then check how recent their work is. Someone who shipped a great project in 2019 and nothing since is in a different spot from someone who ships every month.

Once you find one strong engineer, look at the people who build alongside them. The same projects pull in the same strong people, so the other contributors on those projects are your next search.

Plenty of the people you find this way have a stale LinkedIn or none at all. We wrote about that in how to find engineers who are not on LinkedIn, and our team gave a talk on it at Transform 2026, Cracked Engineers Are Not On LinkedIn.

Appendix: run the search yourself

We rank people with devRank. To build it, we read the code people ship, the projects they contribute to, and the developers they build alongside. We give the Cracked badge to developers in devRank's top tiers.

With this request, you can find engineers in Berlin whose work is like matter-js:

Ask Claude

Use Vamo to find engineers in Berlin who build work like matter-js, and tell me why each one fits.

API request

curl -X POST https://api.vamotalent.ai/v2/developers/search 
  -H "Authorization: Bearer vamo_sk_..." 
  -H "Content-Type: application/json" 
  -d '{
    "rank_by": ["v", "ANN", { "like": ["liabru/matter-js"] }],
    "filters": ["And", [
      ["location_ids", "ContainsAny", { "place": ["Berlin"] }],
      ["is_bot", "Eq", false],
      ["is_org_account", "Eq", false]
    ]]
  }'

You'll find every filter, sort and field you can send in the search API reference.

From Claude, ChatGPT or Cursor over MCP, you can ask in plain words: "find engineers who build work like matter-js, in Berlin, and tell me why each one fits." You can also browse Vamo's Discover pages at /engineers or start from a project at /repos.

FAQ

What does cracked engineer mean? At Vamo, a cracked engineer sits in a GitHub hotspot: a cluster of code, activity and social signal where other strong engineers build, review and star the same projects. Their title often matches the work. We look at where they sit in that cluster and what they shipped there.

Can you screen for cracked engineers? You can screen for the evidence: projects that other people depend on, changes that maintainers merged into hard codebases, and recent, steady shipping. Then in interviews you see how they work on your problems.

Are cracked engineers only in AI? No. We see the same pattern in every kind of engineering, from browser polyfills to physics engines to databases.

How do I find cracked engineers? We look at the work: the repositories people own, the projects they contribute to, and the people they build alongside. With Vamo you can search all three.

Problem with this page? Report it
What is wrong?