AI Safety & Deployment TPM

Slope; San Francisco (CA); Hybrid; USD 160,000 - 230,000; Full time

Benefits offered by this job

Relocation assistance

Hybrid work model (3 days in SF office

Job summary

OpenAI is seeking a Technical Program Manager to streamline safety governance for our ChatGPT, API, and frontier models. Based in San Francisco, CA, this role operates in a hybrid model with relocation assistance available.

You will coordinate risk management and safety review processes to ensure responsible deployment of new models. You will collaborate with research, product, engineering, legal and policy teams to standardize risk assessment lifecycles, develop compute roadmaps, and document

Qualifications

• Advanced degree or equivalent technical leadership experience.
• Proven track record delivering complex technical projects under tight deadlines.
• Strong collaboration with research, product, engineering, legal and policy teams.

Responsibilities

• Track safety research progress and risk tables across programs.
• Oversee human data campaigns' quality for safety systems.
• Develop compute roadmap and resource allocation for safety priorities.
• Coordinate pre-launch safety reviews and post-launch followups.
• Share launch calendars and safety practices with partners (e.g., Microsoft).
• Document safety work with metrics and progress across teams.

Skills

Program management

Safety governance

Stakeholder management

Technical understanding of AI safety

Education

Advanced degree (PhD preferred)

Tools

Risk assessment frameworks

Documentation tooling

Job description

OpenAI is seeking a Technical Program Manager to streamline safety governance for our ChatGPT, API, and frontier models. Based in San Francisco, CA, this role operates in a hybrid model with relocation assistance available.

You will coordinate risk management and safety review processes to ensure responsible deployment of new models. You will collaborate with research, product, engineering, legal and policy teams to standardize risk assessment lifecycles, develop compute roadmaps, and document