Engineering · San Francisco, CA

Research Scientist

This is a research scientist seat for our client, an early-stage company building long-running AI agents that pursue goals over days or months. It's on site in San Francisco. A paid in-person work trial may be part of the process. Visa sponsorship may be available.

What you will do

  • Post-train LLM and multimodal agents with reinforcement learning.
  • Work on long-horizon planning, memory, recovery, and calibrated escalation.
  • Build computer-use and tool-use environments, data pipelines, benchmarks, and evals.
  • Own research bets from end to end.

What we are looking for

  • You have exceptional ML research and engineering experience in post-training, reinforcement learning, agents, reasoning, long-horizon work or memory, computer use, or evals.
  • You have strong experimental judgment.
  • You build complete systems, not only a piece of one.
  • You have a record of significant research ownership: a model, a benchmark, a paper, or open source.
  • You're comfortable at an early-stage company.

Nice to have

  • A publication, open-source work, or a benchmark record.
  • You've built eval harnesses.

Who is hiring

W&W Recruiting is running this search for our client, an early-stage company building long-running AI agents that pursue goals over days or months. We don't put the company name on the page, and we'll walk you through the rest on a call.

Visa sponsorship may be available.

W&W considers every qualified person for this search. We don't screen on age, race, color, religion, sex, national origin, disability, veteran status, or any other characteristic protected by law.

Related: engineering recruiting.