Senior Manager, AI Engineering (Agent OS Platform)

Ready to be a Titan? ServiceTitan is building the Agent OS for the trades: a shared platform that powers role-specific AI experiences across Atlas, field, office, voice, mobile, and future product surfaces. This is not a collection of chatbots. Agent OS is the runtime, context, memory, action, trust, and evaluation layer that lets AI agents help contractors run their businesses safely, observably, and at enterprise scale. We are looking for a Senior Engineering Manager to lead a small, hands-on AI platform team building the core Agent OS. This is a builder-manager role. The right person can lead engineers, shape architecture, make high-quality technical decisions, and stay close enough to the work to unblock design, implementation, debugging, evaluation, and production delivery. This is not a pure people-management, AI strategy, or research leadership role. We need someone who can earn credibility with senior engineers by improving the technical work, not just coordinating it. We do not expect one person to have built every part of an agent platform before. We do expect strong engineering judgment, production scars, hands-on curiosity, and the ability to learn fast while making high-quality technical decisions. What You’ll Build You will lead a compact AI platform engineering team responsible for foundational Agent OS capabilities, including: Agent runtime and workflow execution: role-specific agents, planning, tool use, delegation, pause/resume, long-running workflows, durable checkpoints, and failure recovery. Context and memory systems: retrieval, tenant-aware memory, transcripts, artifacts, tool results, provenance, freshness, and replayable evidence. Capability platform: reusable domain capabilities that combine prompts, tools, context requirements, policies, evals, rollout controls, ownership, and rollback expectations. Action and trust layer: typed action contracts, scoped permissions, business precondition checks, approval flows, reversibility, idempotency, audit trails, and human-in-the-loop controls. Evaluation and observability harness: offline and online evals, scenario libraries, simulation, trajectory review, regression detection, quality metrics, cost/latency telemetry, and autonomy promotion gates. ServiceTitan integration: secure access to systems of record, governed data sources, domain context, Atlas integration, and role-specific agent experiences for owners, CSRs, dispatchers, technicians, managers, accountants, and back-office teams. What You’ll Do Lead the team through architecture, implementation, production launch, and fast iteration. Stay hands-on: review designs and code, inspect traces, debug production behavior, evaluate prototypes, and help engineers make pragmatic tradeoffs. Translate Agent OS strategy into concrete platform slices that ship quickly without creating one-off agent implementations. Define platform contracts for role shells, capabilities, tools, actions, approvals, context, memory, evidence, and evaluation. Build the distinction between what an agent can do and what it is authorized to do in a given tenant, role, workflow state, and risk context. Partner with Product, Design, Architecture, Security, Data Platform, Atlas, and domain engineering teams to create useful, safe, and measurable agent capabilities. Drive evaluation as part of everyday engineering: scenario design, regression suites, trace review, simulation, production monitoring, quality gates, and rollout criteria. Help the team make model and inference tradeoffs across latency, cost, quality, structured outputs, caching, fallback behavior, and provider choices. Ensure live ServiceTitan systems of record remain authoritative while memory, retrieval, transcripts, and agent-generated artifacts are governed as contextual evidence. Work through real agent failures with the team: wrong tool calls, stale context, missing permissions, unsafe actions, poor retrieval, bad recommendations, latency spikes, and cost regressions. Hire, coach, and retain strong engineers who can build fast, reason deeply, and operate responsibly in a fast-moving AI environment. What You’ll Bring 8+ years of software engineering experience, including 4+ years leading engineering teams or major technical initiatives in a product or platform organization. Strong technical background as a builder. You may not write production code every day, but you can read code, review implementation plans, reason through distributed systems, and debug real behavior. Recent hands-on technical leadership: you have personally reviewed design docs, read implementation details, inspected production traces/logs, or debugged system behavior in the last 6–12 months. Experience shipping AI, ML, data, platform, infrastructure, workflow, automation, or developer-platform systems in production. Practical understanding of modern LLM application architecture: model gateways, prompt/context assembly, retrieval, tool calling, structured outputs, memory, agent workflows, and human approval patterns. Strong instincts for production agent safety: typed tools, scoped permissions, business invariants, precondition checks, approval thresholds, reversible actions, idempotency, audit trails, and rollback. Production-minded approach to evaluation: scenario design, behavioral evals, regression suites, trace review, simulation, offline/online metrics, and monitoring for non-deterministic systems. Strong engineering judgment across APIs, distributed systems, event-driven systems, data platforms, observability, reliability, security, and multi-tenant SaaS constraints. Strong data and context instincts: SQL, unstructured data, vector search, metadata, provenance, source authority, freshness, and privacy boundaries. Ability to turn ambiguous strategy into sequenced roadmaps, measurable outcomes, and clear ownership. Clear communication with engineers, product leaders, architects, security