Staff AI Product Manager, Phoenix Platform

AI Proficiency at Sword Health

AI fluency is a core expectation at Sword Health. Every candidate is assessed against our three-level framework — be ready to share real examples of how AI is already part of how you work.

• Explorer (Level 1) — Uses AI daily to boost personal productivity

• Builder (Level 2) — Creates workflows and tools that elevate the whole team

• Integrator (Level 3) — Embeds AI into products and processes at scale

Every hire must demonstrate at least Level 1. The expected level will vary depending on the seniority of the role.

What you’ll do:

Own Sword’s Applied AI platform roadmap, from strategic initiatives and innovation bets through to platform upkeep;
Work at the intersection of AI and product management, hands-on with the agents, evals, harnesses, and observability that our care products are built on;
Own the trade-offs that shape the platform: cost, latency, quality, and when to move to open source models or bring something in-house;
Scope initiatives with Engineering, AI, and Product teams, carefully balancing reusability with differentiation;
Execute by iterating closely with Engineering, Product, Clinical, Operations to guarantee scope adapts and deliverables do not drift from high-value member experiences;
Hold the quality bar, including after release. A model that degrades quietly in production is your problem;
Own safety alongside our clinical team. Our agents talk to people about their bodies and their recovery, and you’ll know how we prevent, detect, and respond when they get something wrong.

What we’re looking for:

5+ years in product management, including recent hands-on ownership of an AI-powered product;
Significant technical depth in AI, generative AI in particular. You understand how models, prompting, retrieval, and agent architectures actually behave, not just what they’re supposed to do;
You’ve put LLM-based features in front of real users and dealt with what came back: the evaluation, the annotation, the observability, the failure modes nobody predicted;
You can hold your own in an AI or engineering discussion, and follow a nuanced technical decision through to what the member will actually experience;
You disagree up and sideways. We expect pushback any time the member experience is being compromised, whatever the source;
Ability to efficiently track the state of the art and, exercising sound product and technical taste, bring innovation;
You write and speak clearly enough to cut through noise. At our size that’s a load-bearing skill, not a nice-to-have;
You get real leverage out of coding agents: debugging, data analysis, production monitoring, small builds, one-off tooling, research, process automation. No slop;
You can manage and grow a small product team.

Bonus points:

Memory, guardrails, or multi-agent orchestration;
Voice AI experiences shipped to production;
Experimentation or evaluation infrastructure you built rather than inherited (A/B testing, evals, human annotation);
Working knowledge of how foundation models are trained and deployed;
Healthcare, or another regulated domain where being wrong has consequences.