AI Data & Evaluation Program Lead

HUD; San Francisco (CA); Hybrid; USD 150,000 - 230,000; Full time

Benefits offered by this job

Medical/dental/vision covered

Office meals

Holiday break

Equinox membership

401k

Commuter benefits

Token access to AI tools

Job summary

HUD is seeking a Technical Program Manager to own complex, time-sensitive data and evaluation programs for frontier AI labs, data vendors, and internal teams. You will turn ambiguous technical asks into executable programs and coordinate the people and dependencies required to deliver them.

You will also manage vendors and build the processes, tooling, and operating rhythms that allow HUD to deliver reliably at increasing scale.

Qualifications

• Experience owning complex technical programs in AI/ML, data ops, or cross-functional settings.
• Strong ability to judge data quality using metrics and samples.
• Proven track record of building and improving data processes and delivery.
• Ability to bring structure to ambiguous problems and manage evolving requirements.

Responsibilities

• Own data and evaluation programs end-to-end from scoping to delivery and retrospective.
• Translate ambiguous requests into clear milestones, owners, and acceptance criteria.
• Maintain data quality procedures using quantitative signals and qualitative inspection.
• Manage external vendors across the delivery lifecycle.
• Improve delivery workflows, documentation, automation, and handoffs.
• Collaborate with research, platform, and marketplace teams to drive product improvements.

Skills

Complex programs

Data ops

Cross-functional

Data quality

Vendor management

Job description

HUD is seeking a Technical Program Manager to own complex, time-sensitive data and evaluation programs for frontier AI labs, data vendors, and internal teams. You will turn ambiguous technical asks into executable programs and coordinate the people and dependencies required to deliver them.

You will also manage vendors and build the processes, tooling, and operating rhythms that allow HUD to deliver reliably at increasing scale.