Research Scientist, Oversight
San Francisco, CAJob$200–600K/yrPosted todayStill listed today
Most applications go out cold — see where you stand first. No sign-up to start.
Watch jobs like this. New roles like this one near San Francisco, CA, by email.
Don't just apply. Show up ready.
Olive works from this exact posting.
At a glance
Olive lists jobs from US employers, including remote roles you can work from the United States.
Job overview
Fleet AI seeks a Research Scientist to develop scalable oversight for agents, detect and explain misbehavior, and keep verification reliable as models become more capable. The role uses Fleet’s sandboxed environments and training tasks as a controlled laboratory, with research findings translated into papers, benchmarks, training gyms, models, or production systems. The scientist will collaborate directly with frontier labs.
Skills & qualifications
Skills
Qualifications
Full job description
About Fleet
Fleet studies how environments produce intelligence: the environment determines what capabilities develop, what behaviors survive, and what "good" looks like. We build the benchmarks that expose where frontier models break and the training that closes the gap. Backed by Sequoia Capital, Menlo Ventures, BCV, and SV Angel.
About the Role
We're looking for a Research Scientist to work on scalable oversight for agents: detecting misbehavior like reward hacking and deception, understanding why it happens, and keeping verification reliable as models outgrow the humans and automated systems checking them. Because Fleet builds sandboxed environments and training tasks for agents, our task factory will be your controlled laboratory where you can intervene, attribute behavior to its cause, and feed what you learn back into improvements. You'll own research threads that ship as papers, benchmarks, training gyms, models optimized for efficient oversight, or production systems, in direct collaboration with frontier labs.
What You'll Work On
-
Scalable oversight: methods that keep verification reliable as tasks get longer and models get stronger.
-
Monitoring: detecting reward hacking or undesirable behavior patterns in our agent traces.
-
Failure mode analysis and attribution: tracing agents' mistakes and misbehaviors upstream to root causes in data, environments, or model weights.
-
Turning oversight into training signal: better verifiers, process supervision, and models trained to catch what humans miss.
What We're Looking For
-
Publication record at top venues (NeurIPS, ICML, ICLR, or equivalent), or equivalent shipped research, in oversight, monitoring, evaluations, alignment, interpretability, AI control, or RL.
-
Strong empirical instincts; comfortable with LLM APIs, harnesses, and large trace datasets.
-
High ownership: you run your own research threads and are measured by what you ship.
Location
San Francisco (on-site).
Compensation
Salary range: $200,000–$600,000 USD annually, plus meaningful equity.
How this job compares
Set against the 2,291 open Research Scientist jobs on Olive Jobs, counted October 29, 2026.
- Pay
- This posting$200–600K/yr
- California median$180–270K/yr497 salary ranges
- US median$150–228K/yr1,544 salary ranges
This range's midpoint, $400K, is 77% above the California median midpoint ($226K) — in the top quarter of Research Scientist salary ranges in California.
- Openings
- Olive Jobs lists 715 open Research Scientist jobs in California and 2,291 across the US. 3% of them nationwide are listed as remote or flexible.
- Experience
- Research Scientist postings that state a minimum ask for 5 years of experience at the median; the middle half ask for 2 to 6 years (1,263 postings).
- Education
- 1,200 of the 2,291 open Research Scientist jobs (52%) ask for a bachelor's degree or higher.
Counted across the open, public US postings on Olive Jobs on October 29, 2026. Pay medians are the middle values of the low ends and of the high ends of postings that list both ends of a range in the same cadence (a yearly salary or an hourly rate); a range's midpoint is halfway between its ends. Pay and experience figures resting on fewer than 20 postings are left out.
You've read the whole posting — now see how you match it.