Principal Product Manager, Agentic AI

Overview

Lead product work for Legion's agentic AI surface, taking AI agents that carry out workforce management work from operational research through specification, prototype, evaluation and enterprise deployment, working hands-on throughout.

What you'll do

• Determine what an agent is responsible for, which tools it may call, what it must escalate, what it must never do, and what a manager sees when it acts.
• Identify which workflows suit an agent and which do not, by spending substantial time inside customer operations and returning with a ranked, defensible view of where an agent creates measurable value.
• Define, for each agent action, where autonomy is appropriate, the escalation thresholds, how a manager overrides the system, what is logged, and how a compliance team audits the outcome.
• Stay engaged through implementation: review the build against the specification, use the product before others are asked to, join bug triage and design review, and hold a release when quality is not there.
• Build and maintain the evaluation sets that determine whether an agent is ready to ship and whether it is degrading in production.
• Write Functional Requirements Documents as the ground-truth specification layer and keep them accurate after release.
• Partner with go-to-market on packaging, metering, pricing inputs, and field readiness.
• Lead agent deployments with large enterprise design partners, including scoping the pilot, establishing success criteria, joining failure reviews, and determining what becomes generally available.

What you'll need

• 7+ years in product management (Senior) or 10+ years (Principal).
• At least 2 years shipping LLM-backed features to external, paying users, not internal tools, proofs of concept, or hackathon projects.
• Demonstrated hands-on capability: writing and iterating on prompts and system prompts, standing up working prototypes with tool definitions, structured outputs, retries and guardrails, and using agentic coding tools (Claude Code or equivalent) daily.
• Ability to review an execution trace and diagnose whether a failure originates in a tool description, a retrieval gap, a context-window constraint, an under-specified output schema, or a genuine model limitation.
• Ability to build and run evaluation sets and interpret the results without a data scientist.
• Informed position on retrieval design, tool-use patterns, structured output, model routing, latency and cost trade-offs, and multi-tenant safety in enterprise SaaS.
• Substantial enterprise B2B SaaS experience: multi-tenant architecture, role-based access control, audit trails, single sign-on, data residency, security review, and enterprise procurement.
• A record of taking an ambiguous, operationally complex domain problem through to a shipped product with demonstrated adoption.
• Evidence of staying with a product through implementation and owning its quality at release, including a release delayed or stopped and how the problem was found.
• Judgment to recognize what current models cannot do reliably, and willingness to say so.
• Comfort working inside a codebase: able to read the code behind a specification, run it locally, and open a pull request. An engineering background is not required.
• Strong written communication.

Nice to have

• Direct experience in workforce management, HCM, ERP, supply chain, or another operationally complex enterprise domain with compliance exposure.
• Familiarity with labor compliance regimes such as predictive scheduling ordinances, meal and rest break rules, and union work rules, or ability to become fluent in a comparable regulatory domain quickly.
• Background in demand forecasting, optimization, or decision systems.
• Experience building platform or authoring surfaces that customers configure themselves.
• An engineering or technical degree, or prior experience as an engineer or data scientist.
• Experience with products sold into frontline-heavy industries: retail, hospitality, healthcare, or manufacturing.

Details

• Individual contributor role with no management responsibility at present.
• Location: Remote, United States.
• Fully remote employer.
• Selection includes a practical exercise: an agent design plus a self-built working prototype using any model provider, discussed with future colleagues. Candidates are compensated for the time.