Senior AI Engineer (Agent OS Platform)

Ready to be a Titan? ServiceTitan runs the businesses behind the trades: jobs, trucks, technicians, equipment, contracts, payments, warranties, compliance obligations, and customer history. That operational context is our advantage. We are building Agent OS to turn that context into safe, observable, production-grade agent work. Agent OS is the shared runtime, context, memory, action, trust, and evaluation layer behind role-specific AI experiences across Atlas, office, field, voice, mobile, and future product surfaces. This is not a collection of chatbots. It is the platform that lets agents help contractors run their businesses with the right evidence, permissions, approvals, and audit trails. You will help build the core engineering primitives behind that platform: agent runtime, typed tools, context and memory assembly, trust and approval flows, evaluation infrastructure, and production observability. You are not building one agent for one product surface. You are building the platform that product teams use to build many agents safely. You will work on a small, senior AI platform team and partner closely with Product, Architecture, Security, Data Platform, Atlas, and domain engineering teams. What You’ll Build Agent runtime and workflow execution: Build the runtime for role-specific agents, tool use, delegation, pause/resume, durable checkpoints, retries, and failure recovery. Agents must resume safely without losing state or duplicating side effects. Typed tools and action contracts: Build deterministic controls around non-deterministic reasoning: governed reads, proposed writes, precondition checks, business invariants, scoped permissions, idempotency, audit trails, and rollback. Context and memory systems: Build tenant-scoped context assembly, retrieval, freshness controls, provenance, transcripts, artifacts, tool results, and replayable evidence. ServiceTitan systems of record stay authoritative; memory provides context and coordination. Trust and approval infrastructure: Build human-in-the-loop gates, approval thresholds, reversibility, tenant policy enforcement, and audit history for financial, contractual, dispatch, warranty, and compliance-sensitive workflows. Evaluation and observability: Build offline and online evals, scenario libraries, simulation, trajectory review, regression detection, cost and latency telemetry, and autonomy promotion gates. Reusable capability platform: Help product teams package prompts, tools, context requirements, policies, evals, rollout controls, ownership, and rollback into governed capabilities for owners, CSRs, dispatchers, technicians, managers, and back-office teams. Model and inference architecture: Make practical tradeoffs across latency, cost, quality, structured outputs, caching, fallback behavior, provider choice, and model routing behind a shared platform layer. What You’ll Do Design and implement core Agent OS platform services. Write production code and review implementation details from other engineers. Build reliable APIs, workflows, tools, and services for agent execution. Inspect traces, debug failures, and improve production behavior. Design evaluation scenarios and regression suites for agent workflows. Work through real agent failure modes: stale context, wrong tool calls, missing permissions, unsafe actions, poor retrieval, latency spikes, and cost regressions. Partner with domain teams to turn agent use cases into reusable platform patterns. Help define platform contracts for tools, actions, approvals, context, memory, evidence, and evaluation. Contribute to technical direction while staying grounded in what can ship quickly and safely. Communicate clearly with engineers, product managers, architects, security partners, and engineering leadership. What You’ll Bring 5 years of production software engineering experience. Strong hands-on coding ability in Python, Java, C#, or another backend language. Python experience is strongly preferred ... (truncated, view full listing at source)