SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 280,000 - 300,000 / annual
Legion is building AI agents that execute workforce management work—not merely describe it. The company is solving the substantial problem of software that reasons over demand, labor regulation, and employee preferences, then acts with human oversight where stakes require it. This Principal Product Manager role leads the product work to take an agent surface from operational research through specification, prototype, evaluation, and enterprise deployment, working hands-on throughout.
Key responsibilities include:
**Agent Definition**: Determine agent scope, tools, escalation thresholds, and what managers see when agents act. These decisions directly affect employee pay and legal compliance.
**Problem Selection**: Spend substantial time inside customer operations (district managers, store managers, labor operations, frontline employees) to identify which workflows suit agents and which do not, returning with a ranked, defensible view of where agents create measurable value.
**Human-in-the-Loop Contract**: Define autonomy boundaries, escalation thresholds, manager override mechanisms, logging, and compliance audit trails. This is treated as a first-class design problem.
**Build Quality**: Stay engaged through implementation—review builds against specification, use the product before others, join bug triage and design review, catch drift early, and hold releases when quality is insufficient. You are accountable for what ships, not what was specified.
**Evaluation**: Build and maintain evaluation sets that determine agent readiness and production degradation. Write test cases, review failure traces, set pass criteria, and hold releases when evaluation does not support demonstration.
**Specifications**: Write Functional Requirements Documents as ground-truth specifications. Release notes, enablement material, and sales content are generated from maintained specs rather than reconstructed from code.
**Commercialization**: Partner with go-to-market on packaging, metering, pricing, and field readiness. Hold a considered position on difficult commercial questions (unit metering, what customers pay for when software performs work a person previously did).
**Design-Partner Delivery**: Lead agent deployments with large enterprise design partners—scope pilots, establish success criteria, join failure reviews, determine what becomes generally available versus bespoke.
**Technical Expectations**: This role requires hands-on building. In a typical week: write and iterate on prompts and system prompts directly; stand up working prototypes of proposed agents (tool definitions, structured outputs, retries, guardrails) before engineering builds; use agentic coding tools (Claude Code or equivalent) daily for prototyping, data analysis, specification work, and reading codebases; review execution traces to diagnose failures; build and run evaluation sets; hold informed positions on retrieval design, tool-use patterns, structured output, model routing, latency/cost trade-offs, and multi-tenant safety; open pull requests against the codebase (permitted and encouraged but not required); follow model releases, agent frameworks, and protocol changes, translating them into roadmap implications within days.
This is an individual contributor role with no team to inherit and no management responsibility at present. Direction is not handed down in detail; the environment carries genuine ambiguity. The category is being defined as the work progresses, and model capabilities shift underneath the roadmap.
**REQUIREMENTS**
- 7+ years in product management (Senior level) or 10+ years (Principal level), including 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 as described above; expect to show your work
- Substantial enterprise B2B SaaS experience: multi-tenant architecture, role-based access control, audit trails, single sign-on, data residency, security review, and enterprise procurement. Should anticipate what a customer's security questionnaire will ask about an LLM feature
- 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 (will be asked about a release you delayed or stopped, what you found, and how you found it)
- Judgment to recognize what current models cannot do reliably, and willingness to say so when a room expects a different answer
- Comfort working inside a codebase. Engineering background not required, but should be able to read code behind specifications, run it locally, and open a pull request
- Strong written communication. Legion runs on documents; candidates evaluated substantially on how they write
**PREFERRED QUALIFICATIONS**
- Direct experience in workforce management, HCM, ERP, supply chain, or another operationally complex enterprise domain with compliance exposure
- Familiarity with labor compliance regimes (predictive scheduling ordinances, meal and rest break rules, union work rules) or demonstrated 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
- 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
**SELECTION PROCESS**
The process includes a practical exercise: candidates produce an agent design (scope, tools, failure modes, human-in-the-loop contract, evaluation plan) together with a working prototype built themselves using any model provider. The exercise is discussed with people the candidate would work alongside. Candidates are compensated for the time.
