Owner of a specific portfolio of AI products or transformation projects, accountable for landing them on time, on budget, and against measurable ROI targets. By
An AI Program Director delivers complex AI programs: coordinating engineering, data, legal, security, and business workstreams across multiple projects with shared dependencies, budgets, and executive stakeholders. It's senior program management craft applied to AI's special physics: probabilistic outcomes, eval-gated releases, regulatory checkpoints, and fast-moving vendor dependencies.
In 2026 the role is the connective tissue of enterprise AI delivery: translating between executives and technical leads, keeping governance artifacts audit-ready, managing model/vendor changes mid-flight, and protecting programs from both hype-driven scope creep and risk-driven paralysis.
Outcomes are probabilistic and gated by evals rather than fixed acceptance tests; releases pass model-risk and compliance checkpoints; vendor dependencies shift mid-flight as models update. The craft is classic program discipline plus AI-specific gate design.
Roughly $160k–$250k base in the US, with regulated-industry and revenue-critical programs at the top of the band.
Conversancy, not engineering depth: you must understand RAG, agents, evals, and LLM operations well enough to challenge estimates and spot risk. Directors without that literacy lose the room with technical leads.
A value-tracked roadmap, eval gates with pass criteria per release, an AI system inventory with risk classifications, audit-ready documentation, and vendor-change contingency plans. Auditors and executives increasingly ask for exactly these.