Job Description
Job DescriptionWhat you'll own\n
- \n
- \n
Evaluation suites built from production traces: deciding what to measure, building the graders, and proving they track real quality.
- \n
Verifiers and reward models. Turning human corrections, outcomes, and standards of good work into training signal.
- \n
Post-training experiments end-to-end: hypothesis, data, SFT/DPO/RL runs, ablations, analysis, and the write-up that defends the win with numbers.
- \n
RL environments and agentic harnesses for the workflows we're specializing models for.
- \n
Reproducing recent post-training, reasoning, and agent papers, and translating the ones that work into our stack.
\n
\n
\n
\n
\n
About you\n
- \n
- \n
Currently a student (BS / MS / PhD) in CS, ML, or an adjacent field, or recently graduated.
- \n
Strong empirical ML background. You've run post-training or RL experiments on LLMs and can explain what worked, what didn't, and why.
- \n
Rigorous about evaluation. You can spot a benchmark that lies and design one that doesn't.
- \n
Comfortable in messy production data. Real traces, not curated datasets, are your raw material.
- \n
Bias to ship. Your experiments end in a deployed model or a killed hypothesis, not just a doc.
- \n
Available for at least 12 weeks at meaningful intensity.
\n
\n
\n
\n
\n
\n
Bonus\n
- \n
- \n
Publications or OSS in post-training, RL, evals, or agents.
- \n
Experience with verifier models, synthetic data, or RL environments at scale.
- \n
Shipped research into a production system before.
- \n
Strong taste for what the field is over- vs. under-investing in.
\n
\n
\n
\n
