Staff AI Engineer $225K–$250K + meaningful equity San Francisco, CA (Onsite) A well-funded AI infrastructure startup is hiring a Staff AI Engineer to help build the core agentic intelligence layer powering automation inside complex engineering software environments. The product already has meaningful traction with Fortune 100 customers and is backed by top-tier investors in the AI ecosystem. This is a highly technical, deeply hands-on role for someone who wants to work on difficult real-world agent problems, not lightweight chatbot wrappers or internal prototypes. The Opportunity You’ll own foundational agent architecture and help define how AI systems reliably execute complex workflows inside real enterprise environments. This role sits directly alongside company leadership and will heavily influence the future direction of the platform, evaluation infrastructure, and broader AI engineering strategy. The environment is highly execution-oriented. Leadership is measured through technical contribution, system ownership, and shipping production systems under ambiguity. What You’ll Be Doing • Build and improve production agentic AI systems capable of executing multi-step workflows across desktop software environments • Own core architectural decisions around: • Tool orchestration • Context management • State handling • Error recovery • Model routing • Workflow execution reliability • Design and scale evaluation frameworks measuring: • Workflow success rate • Reliability • Failure modes • Cost efficiency • Regression detection • Define token budgets and optimize inference efficiency for commercially viable agent execution • Work closely with researchers, domain experts, and product stakeholders to translate real user workflows into measurable agent benchmarks • Lead technically while remaining highly hands-on in implementation and architecture • Collaborate with customers and internal teams to improve workflow coverage and production performance What They’re Looking For • Strong production experience building agentic AI systems, not just LLM-powered interfaces • Deep understanding of: • Tool-calling agents • Multi-step orchestration • State management • Context handling • Workflow reliability under ambiguity • Strong Python engineering background • Experience designing evaluation and benchmarking systems for AI agents or complex ML systems • Comfort operating in highly ambiguous startup environments with broad ownership • Strong systems thinking and architecture instincts • Ability to remain deeply hands-on technically while influencing engineering direction Strong Signals • Experience with: • SWE-bench, GAIA, or similar evaluation frameworks • LangSmith, Logfire, tracing, or observability tooling • Agent orchestration frameworks • Desktop automation • Enterprise AI deployments • Workflow automation systems • Startup experience is highly valued, especially in technically demanding product environments Location / Work Setup • Full-time onsite in San Francisco • Flexible hours and a high-autonomy environment • Compensation includes meaningful equity participation