All roles
Machine Learning Engineer
You will make a coding agent better at the work people actually give it: long tasks, unfamiliar codebases, and tools that fail in ordinary ways.
What you'll do
- Design evaluations that measure what matters: task completion, tool use, latency, cost, and regressions.
- Improve context selection and memory: what the agent sees, keeps, and retrieves across long sessions.
- Tune model routing, prompts, and tool schemas across frontier providers, balancing quality against cost.
- Turn production traces and user feedback into datasets, experiments, and changes that ship.
- Work with product and engineering to monitor changes after release, not just in a notebook.
What we're looking for
- Have shipped applied ML or AI systems to production, not only trained models in isolation.
- Strong Python, and comfortable with APIs, data pipelines, and evaluation infrastructure.
- Practical knowledge of LLMs: prompting, tool calling, retrieval, fine-tuning, and evals.
- An empirical streak. You can design an experiment, read the result, and call it.
- Clear writing and comfort working async across time zones.
Nice to have
Coding agents or developer tools, reinforcement learning, inference optimization, TypeScript, open-source work.