Join the team measuring what AI actually does in production.
GOAT pays developers for the agent sessions they already generate and turns them into research built in the open — a read-only capture network, a default-DENY redaction pipeline, and a live real-money benchmark.
A small team with
outsized leverage.
We hire for taste, intensity, and whether you've built things that mattered to you — not pedigree. Every role here owns a real slice of the pipeline, from capture hooks to the redaction gate to the live benchmark.
Any Department
1 roleOpen Application
Don't see a role that fits? We're a small team building the whole pipeline — a capture CLI, encrypted ingest, the redaction gate, payouts, and a live real-money benchmark — and we hire exceptional people ahead of named roles. Send your best work: a project, a repo, a paper, anything that shows how you think and what you want to own.
Engineering
2 rolesMember of Technical Staff, Platform
TypeScript/Fastify ingest, ClickHouse at trace scale, BullMQ pipelines, a Next.js dashboard — plus a Python redaction sidecar. You'll own the pipeline that turns raw agent sessions into research-grade data: encryption at ingest, deterministic IDs and content fingerprints, provider-response attestation, and payouts backed by an append-only audit ledger.
Member of Technical Staff, ML Research Engineer
Own the redaction gate (Presidio + spaCy + deterministic secret detectors), contribution scoring, and the evaluation work behind PolymarketBench — Brier, log-loss, and calibration over real-money forecasts. Your recall numbers decide what's eligible for research, and your evaluation work ships in the open.
Growth Design
1 roleFounding Launch Designer
Own how GOAT shows up in public. PolymarketBench publishes every real-money bet, agent transcript, and score live — turn work like that into launches people can't ignore: product films, benchmark reveals, brand systems for new research artifacts, and distribution across X, LinkedIn, and YouTube. Ship in days, not quarters, with the founders in the room.
What we
look for.
We don't care about pedigree. We care about taste, intensity, and whether you've built things that mattered to you. This is a small team building the pipeline that production-AI research runs on — everyone here has outsized influence, and we're looking for people who are drawn to that kind of leverage and can handle it.
If you've shipped products people actually use, done research that changed how you think, or made something beautiful just because you thought it should exist — we want to talk.
Intro call
20–30 minWe read every application ourselves. If there's a fit, we set up a straightforward call about what you're working on and what we're building. Ask us the hard questions.
Technical deep-dive
60–90 minWe go deep on something you've built, then work through a real problem from our stack together. No gotchas, no puzzles — we want to see how you reason.
Work session
~1 dayYou build alongside the team on a real task from the backlog, so both sides can see what working together is actually like.
Offer
days, not weeksWe move fast and tell you where you stand. If it's a yes, you get a decision — and a concrete first project — within days.
Don't see the right role? Reach out anyway.
We're always looking for exceptional people. Send us your work — a project, a repo, a portfolio, anything that shows how you think.