Technical Program Manager, AI Transformation

Technical background you'll need

Software delivery — deep working grasp of the SDLC end to end (requirements → design → implementation → review → QA → CI/CD → release → maintenance) and how AI changes each stage; version control, PR review, test automation, CI/CD — enough to know what an AI-SDLC audit measures and what "good" looks like.

How AI and LLMs work — conceptual grasp at a working level: tokens and context windows, prompting and in-context learning, why models hallucinate and where they're unreliable, determinism and evaluation. Know real capabilities and limits; you don't train models.

Agentic systems and AI-SDLC tooling — tool/function calling, RAG, orchestration, MCP, guardrails and evals; the AI coding assistants engineers use daily. Enough to tell genuine acceleration from risk and hold your own when an engineer says "the model can't do that."

Must-have -

• Ran complex multi-team programs end-to-end as executor
• Deep SDLC + modern-eng-practice understanding \
• Solid conceptual grasp of AI/LLMs/agentic systems
• Execution discipline (metrics/tracking/deadlines that stick)
• Proven change management through non-reports
• Structural thinker who turns direction into OKRs
• Strong negotiation across engineering/delivery/leadership - hands-on with modern AI tooling, recent and real

• Mid-to-large software-services/consulting background
• Prior engineering/tech-lead/architect role
• Stood up an audit/assessment program across teams
• Hands-on with Cursor/Claude/Copilot/MCP/RAG
• Travel/hospitality domain
• Recent demonstrated AI-engineering learning.

• Own a defined set of AI programs end to end — from framing the objective to a measurable, demonstrable result.
• For each: define the outcome, break it into OKRs, sequence the work, assign owners, set cadence, track to done. -
• Drive AI-native SDLC adoption inside your programs — audit current practice, design rollout, prove real behavior change (engineers working differently, not licenses handed out).
• Design and run the operating rhythm: status, weekly roll-ups, risk/blocker surfacing, deadlines that hold.
• Coordinate the people your programs depend on — engineers, solution architects, auditors, delivery leads — most of whom don't report to you.
• Triage technical resistance. Resolve the routine ~90% yourself with a credible technical answer; escalate only the rare hard architecture calls.
• Feed portfolio-level signal upward: what's working, what's stuck, where a direction-level decision is needed.
• Keep programs honest — separate real adoption from theater, and measure the difference.