Director - Harness Engineering
Core
Design and build the 'harness' environment that steers AI coding agents toward correct, maintainable, well-architected output, enforcing quality by system rather than human audit.
Role type
Director, hands-on IC leading a regional team of harness engineers
Builds
Feedforward guides (agent instructions, skills, architectural rules), feedback sensors (custom linters, verification loops, LLM-as-judge reviewers), and AI governance controls
Domain
AI-assisted software engineering / Agent orchestration / Engineering tooling
Deliverable
production ML models | product features | infrastructure
Required skills
Software architecture, AI coding agents (Claude Code, Codex), engineering tooling (linters, CI/CD, observability), spec-driven development, AI governance, distributed team management, security guardrails for autonomous agents
Preferred skills
Experience with AGENTS.md conventions, fitness functions, deterministic vs inferential controls, cost/speed/reliability trade-offs in agent checks
Technologies
LLMs (Claude, Codex), CI/CD pipelines, containerized environments, custom linters, LLM-as-judge frameworks
Responsibilities
Design and evolve harness components (guides, guardrails, context); build feedback sensors and verification loops; define AI governance boundaries and testing infrastructure; establish cross-organizational QA standards; manage and coach a geographically distributed team; define observability metrics for agent work
Seniority
Director, hands-on IC with team leadership
