Researcher, Misalignment Monitoring Lead
OpenAISan FranciscoOn-Site
Needs a free Superjobs account. You'll finish on OpenAI's Ashby page.
Skill score
Summary
This person leads research and technical direction for evaluating and improving misalignment monitors on frontier AI systems, designing rigorous evaluations and stress-testing monitors for false positives and evasion. It suits a senior researcher with strong ML and AI safety background who can independently drive open-ended problems to measurable results.
- Estimated salary
- Est. $150k–224k/yr
- Estimated for a manager role, ~7 yrs experience (typical for the level), United States. Not published by the employer.
- Work mode and location
- On-Site, San Francisco
- Employment type
- Full-time
Job description
Overview
We are looking for a senior researcher to lead technical work on misalignment monitoring for frontier AI systems. You will develop methods to evaluate the effectiveness of AI monitors, understand their limitations, and improve our ability to detect and stop concerning model behavior reliably.
This role combines research leadership with hands-on technical execution. You will set research priorities, design rigorous evaluations, investigate monitoring failures, and develop approaches for stress-testing and improving monitoring systems. A key focus will be understanding where monitors produce false positives, miss important behaviors, or become less effective as models grow more capable.
You will work closely with researchers and engineers across OpenAI to turn monitoring research into practical improvements to our safety systems. We are looking for someone with strong research judgment, technical depth, and the ability to independently drive ambiguous problems from initial investigation through measurable results.
Roles and responsibilities
- Lead the research and technical direction for evaluating and improving the misalignment monitoring of frontier AI systems.
- Design and execute rigorous evaluations that measure monitor effectiveness, identify failure modes, and assess the reliability of monitoring signals.
- Develop methods to stress-test and red-team monitors, including identifying behaviors that evade detection or expose weaknesses in existing approaches.
- Investigate false positives, missed detections, and other monitoring failures to improve the accuracy and usefulness of safety evaluations.
- Build scalable evaluation frameworks and experimental tools that enable ongoing measurement of monitoring performance as model capabilities evolve.
- Translate research findings into actionable improvements, working with partner teams to strengthen monitoring systems and inform broader safety decisions.
- Establish research priorities and guide technical execution, balancing foundational research questions with practical deployment needs.
Skills required
Top skills
- machine learning research
- model evaluations
- AI safety
- alignment
All skills (8)
- robustness
- experimental design
- large language models
- red-teaming
- adversarial testing
- AI monitoring
- scalable oversight
- interpretability
Qualifications
- Have a strong background in machine learning research, particularly in model evaluations, AI safety, alignment, robustness, or related areas.
- Have demonstrated the ability to lead technically challenging research projects, from identifying important questions through designing experiments and delivering results.
- Bring strong experimental judgment and understand how to design evaluations that are rigorous, reproducible, and difficult to game.
- Have hands-on experience building or evaluating machine learning systems, including working with large language models or other frontier AI systems.
- Enjoy investigating complex failure modes and developing methods to distinguish meaningful safety signals from misleading measurements.
- Are comfortable working on open-ended research problems where the right methodology or success criteria may not be obvious at the outset.
- Combine strong research instincts with practical engineering ability and can independently build experiments, analyze results, and iterate quickly.
- Communicate technical findings clearly and enjoy collaborating across research, engineering, and safety teams.
- Are motivated by OpenAI's mission and the challenge of developing reliable oversight mechanisms for increasingly capable AI systems.
Nice to have
- Experience with AI monitoring, automated evaluations, adversarial testing, or red-teaming.
- Research experience in scalable oversight, AI control, model behavior analysis, or interpretability.
- Experience developing monitoring or evaluation systems that are used in real-world deployments.
- A track record of technical research leadership, including mentoring researchers or coordinating cross-functional research initiatives.
Company info
About OpenAI
OpenAI is an AI research and deployment company that builds AI systems and deploys them to the world through its products. Its customers and businesses use these AI products across their operations, with support from its sales, solutions, and support teams.