Technical Program Manager, Multimodal
About this role
The ChatGPT Multimodal team at OpenAI builds voice, image generation, and other multimodal capabilities, turning frontier research into reliable products for a global user base. Operating out of the San Francisco office, the technical program manager establishes the flywheel enabling multimodal systems to learn from real-world usage and iterate rapidly. Day-to-day responsibilities involve bridging product engineering, research, inference, and capacity teams to close parity gaps, manage resource constraints, and coordinate complex launches.
Candidates joining this team will thrive if they possess deep technical expertise in machine learning infrastructure and a talent for turning fragmented manual processes into scalable operating mechanisms. Success requires navigating ambiguity, communicating clearly with diverse stakeholders, and driving structured debugging without direct authority. Individuals passionate about expanding responsible access to advanced artificial intelligence will find a natural home in this environment.
What you will do
- Design and implement systems to mine production conversations and usage signals to isolate key multimodal workflows and failure modes.
- Manage and maintain evaluation frameworks for priority multimodal behaviors with established quality standards and clear ownership.
- Translate raw production metrics into decision-ready data that research teams utilize to enhance model performance.
- Drive parity between research and production environments by addressing differences in system prompts, sampling behaviors, and inference configurations.
- Forecast demand and coordinate GPU and serving capacity planning for voice and image-generation workloads.
- Orchestrate cross-functional initiatives spanning multilingual data collection, actor recruitment, and external product partnerships.
What you need
- Demonstrated experience leading technically complex programs across machine learning, multimodal products, model evaluation, or inference infrastructure.
- Proven ability to bridge user-facing product behavior with underlying model configurations, serving constraints, and capacity systems.
- Track record of converting production data into prioritized evaluations and measurable engineering improvements.
- Capacity to reason credibly about GPU demand, serving latency, reliability, and launch tradeoffs.
- Experience driving structured debugging and closing parity gaps across teams with distinct operating environments.
- History of building scalable tooling and durable operating practices from manual or fragmented workflows.
Benefits
- Offers Equity
- Relocation assistance
Skills and tech stack
Questions candidates ask
What is the remote work policy for this role?
This role is based in San Francisco, CA. The team uses a hybrid work model requiring three days in the office per week.
What is the compensation range and benefits structure?
The salary ranges from $207K to $445K and includes equity offers. Relocation assistance is also provided to new employees.
What teams will I collaborate with regularly?
You will work closely with product engineering, research, Human Data, inference and capacity teams, safety partners, and external vendors or product partners.
What specific technical areas does this program management role cover?
The work spans production-signal mining, evaluation and data pipelines, research-to-production parity, and multimodal capacity planning for voice and image-generation launches.