Applied Research Intern

What you'll do

• Build evaluation suites and graders from production traces, and validate that they measure real quality.
• Develop verifiers and reward models, and run end-to-end post-training experiments including analysis and write-ups.
• Build RL environments and agentic harnesses, and reproduce relevant research papers for use in the company’s stack.

What they ask for

• Currently a student in CS, ML, or an adjacent field, or recently graduated.
• Strong empirical ML background, including experience running post-training or RL experiments on LLMs.
• Rigorous evaluation skills and comfort working with messy production data.
• Available for at least 12 weeks at meaningful intensity.

Nice to have

• Publications or open-source work in post-training, RL, evaluations, or agents.
• Experience with verifier models, synthetic data, large-scale RL environments, or shipping research to production.

Tools

LLMs, SFT, DPO, RL

AI focus

Post-train and evaluate models for specialized workflows, including reward modeling, verifiers, and agentic systems.