10 OpenAI AI safety researchers
Reward hacking, adversarial robustness, interpretability, scalable oversight and privacy are among the problems these OpenAI AI safety researchers name in their own descriptions of their work, and nine of the ten are still at OpenAI. Teams testing a model for risky behavior, building oversight or red-teaming methods, or deciding which safety research to fund can find someone here whose problem matches theirs.
Professionals to explore
Aleksandar M.
LinkedInExperience: Member of Technical Staff · OpenAI
$1,683 per 30-minute call
Interpretability research is Aleksandar M.'s work as Member of Technical Staff at OpenAI since January 2025. Aleksandar was also Visiting Scientist at Guide Labs (2024-2025) and Independent Researcher at MATS Research (2023-2024), so questions on reading what a model has learned internally are the natural fit.
Is this you? Claim your profile and get paid for calls.
Alex Beutel
LinkedInExperience: Member of Technical Staff, Research Scientist · OpenAI
$4,039 per 30-minute call
Alex Beutel leads OpenAI's Safety Research teams, spanning safety training, robustness, oversight, model policy and interpretability, as Member of Technical Staff, Research Scientist since June 2023. From New York, Alex was Senior Staff Research Scientist (2021-2023), Staff Research Scientist (2019-2021), Senior Research Scientist (2017-2019) and Research Scientist (2016-2017) at Google.
Is this you? Claim your profile and get paid for calls.
Christopher A. Choquette-Choo
LinkedInExperience: Member of Technical Staff (Research Scientist) · OpenAI
$3,365 per 30-minute call
Security and privacy sit inside Christopher A. Choquette-Choo's alignment work as Member of Technical Staff (Research Scientist) at OpenAI since August 2025. Christopher was Senior Research Scientist (2024-2025) and Research Scientist (2023-2024) at Google DeepMind and held two Machine Learning Researcher roles at Google between 2020 and 2023, a fit for privacy and model security questions.
Is this you? Claim your profile and get paid for calls.
Gabriel Wu
LinkedInExperience: Member of Technical Staff · OpenAI
$1,683 per 30-minute call
Gabriel Wu does alignment research as Member of Technical Staff at OpenAI since June 2025 and has been Visiting Researcher at Alignment Research Center since 2023. Gabriel was also Director at AI Safety Student Team (2023-2024) and Research Staff at Redwood Research (2023), which suits questions on how safety researchers are trained and recruited.
Is this you? Claim your profile and get paid for calls.
Juan Felipe Cerón Uribe
LinkedInExperience: AI Alignment Research Engineer · OpenAI
$2,356 per 30-minute call
Adversarial robustness is the stated focus of Juan Felipe Cerón Uribe, AI Alignment Research Engineer at OpenAI since May 2022. Juan was Machine Learning Engineer at Factored (2019-2022) and Research Intern at Stanford Existential Risks Initiative (2021), and describes working to mitigate catastrophic risks from misaligned AGI, a fit for jailbreak and attack-resistance questions.
Is this you? Claim your profile and get paid for calls.
Marcus Williams
LinkedInExperience: Member of Technical Staff · OpenAI
$1,683 per 30-minute call
Based in Sweden, Marcus Williams has been Member of Technical Staff at OpenAI since May 2025, working on deception and reward hacking monitoring and mitigation. Marcus held Research Scholar at MATS Research twice in 2024 and 2025 and was AI Alignment Researcher (LTFF grant) as an independent (2023-2024) on a Long-Term Future Fund project, so reward design questions fit here.
Is this you? Claim your profile and get paid for calls.
Mikita Balesni
LinkedInExperience: Member of Technical Staff · OpenAI
$2,019 per 30-minute call
Member of Technical Staff at OpenAI since December 2025, Mikita Balesni works on AI alignment and evaluations there. Mikita was Research Scientist at Apollo Research (2023-2025) and ML Safety Research Scholar at Stanford Existential Risks Initiative (2022-2023), describes being a founding member at an AI evaluations research organization, and is a fit for questions on evaluating frontier models for misalignment.
Is this you? Claim your profile and get paid for calls.
Ollie Matthews
LinkedInExperience: Member of Technical Staff · OpenAI
$1,683 per 30-minute call
Ollie Matthews, in London, has been Member of Technical Staff at OpenAI since October 2025 and describes the role as AI safety researcher on the Alignment team. Ollie was Research Engineer at AI Security Institute (AISI) (2024-2025) and AI Safety Researcher at LASR Labs (2024), a fit for comparing alignment research at a lab with safety evaluation outside one.
Is this you? Claim your profile and get paid for calls.
Rai (Michael) Pokorny
LinkedInExperience: Member of Technical Staff · OpenAI
$1,683 per 30-minute call
Rai (Michael) Pokorny has been Member of Technical Staff at OpenAI since November 2022, having held Member of Technical Staff (Resident) there in 2022, with the Superalignment team (scalable oversight) and the Reinforcement learning team listed. Rai was Software Engineer at Google (2017-2021), useful for questions on overseeing models that are hard for people to check directly.
Is this you? Claim your profile and get paid for calls.
Ryan Lowe
LinkedInExperience: Former Member Of Technical Staff · OpenAI
$2,019 per 30-minute call
Ryan Lowe was Member Of Technical Staff at OpenAI from 2019 to 2024 and describes research on the safety team studying reward modeling; an OpenAI Research Intern stint (2017) also appears. Ryan has been Research Ecosystem Lead at Meaning Alignment Institute since 2025, and the McGill University PhD work covered multi-agent reinforcement learning and dialogue systems.
Is this you? Claim your profile and get paid for calls.
Choose the right perspective
Alex Beutel is the broadest choice, since the record describes leading OpenAI's Safety Research teams across several areas. For a narrow technical question, match the problem: Marcus Williams for reward hacking, Juan Felipe Cerón Uribe for adversarial robustness, Aleksandar M. for interpretability, Rai (Michael) Pokorny for scalable oversight and Christopher A. Choquette-Choo for security and privacy. Ryan Lowe left OpenAI in 2024, so ask about reward modeling history rather than current practice.
Questions to take into the conversation
- 01How do OpenAI safety researchers detect reward hacking or deceptive behavior during training before a model ships?
- 02Which interpretability or scalable oversight results from OpenAI's alignment work actually change how a model is trained?
- 03How does an alignment team at OpenAI decide which evaluations must pass before a release, and who can block it?
About this directory
This is a professional research starting point based on business profile data retrieved on . Titles and companies reflect that source snapshot; roles marked former had ended by then. Check the linked profiles for current details. Inclusion does not imply Instant Expert membership or availability.