Senior AI Product Manager- EZRA

Ezra•New York, NY

9d•Hybrid

About The Position

EZRA is seeking a Senior AI Product Manager to own the AI experience within its coaching platform, including Cai. AI is central to the product, and the role is accountable for the user experience, its effectiveness, and the economics of operating it. The Product Manager partners with Engineering, which owns technical architecture and model selection, and with coaching experts, who provide the coaching methodology. Success means becoming the clear product owner for the AI experience: making independent, high-quality product decisions and building AI coaching experiences that users trust, measurably improve outcomes, and make economic sense at enterprise scale. The role spans product direction, experimentation, evaluation, launch, and iteration, with particular attention to trust, confidentiality, quality, latency, reliability, safety, and cost.

Requirements

Significant product management experience in AI-native or AI-first products.
Personal experience shaping and shipping an LLM-based conversational product designed to support behavior change.
Practical understanding of modern LLMs, including their limitations, failure modes, and evaluation.
Experience reviewing sensitive conversational data at scale within privacy controls and translating findings into product decisions.
Ability to build prototypes using code, AI-assisted development tools, or low-code environments.
Startup or scale-up background with a track record of measuring outcomes, not just usage.

Nice To Haves

Experience in digital mental health, health coaching, AI tutoring, AI coaching, or another behavior-change product.

Responsibilities

Own AI capabilities within the coaching platform, including Cai, from product direction through launch and iteration.
Decide what the AI should do, how users experience it, how its effectiveness is measured, and what it costs to run.
Turn emerging LLM capabilities into product experiences that create meaningful, measurable behavior change.
Lead rapid experimentation across prompts, context, memory, retrieval, agents, and tools; prototype ideas when that is the fastest way to learn.
Define evaluation methods for probabilistic AI experiences using conversational data review, product data, and outcome measures—not engagement alone.
Manage trade-offs across quality, latency, reliability, safety, and cost, while designing for trust and confidentiality in highly personal conversations.