Senior AI Product Engineer

About Health Note

Health Note is building AI-powered digital assistants that act like trusted members of the care team. Instead of rigid, disconnected software modules, our assistants gain skills over time — from handling simple reschedules to managing complex intake, referrals, medication refills, and patient access workflows.

Patients experience a natural, empathetic interaction while providers gain a system that is reliable, transparent, and continually improving. Our goal is to deliver a concierge experience for every patient and a scalable path to automation for every clinic.

About the Role

We're looking for a product-minded engineer who can make AI systems measurably better in production. This is a product engineering role first: you'll write production code, build internal and product-facing systems, and use measurement to decide what to improve next.

As a Senior AI Product Engineer, you'll build systems that help us understand how Health Note's Access Agent is performing, where it breaks down, and which changes actually improve outcomes. You'll connect agent behavior, customer workflows, prompts, configurations, and outcome data so we can improve the system with confidence.

Success is measured by whether the systems you build make the Access Agent more reliable, measurable, and effective in production.

This is not a pure ML research role or a dedicated evals-platform role. You'll build AI product systems, use measurement to decide what to improve next, and create enough feedback structure to make changes safely.

This role is ideal for an engineer who likes ambiguous product problems, trusts data over assumptions, and can turn messy real-world behavior into systems for measurement and improvement.

Responsibilities

• Own improvements to the Access Agent from investigation through production rollout, measurement, and follow-up.

• Build practical ways to measure agent performance, including evals, dashboards, review workflows, logs, and production metrics.

• Analyze transfer patterns, containment rates, task success, and operational outcomes to identify improvement opportunities.

• Build tooling and workflows that connect prompts, configurations, interventions, customer workflows, and outcomes.

• Create checks that help us catch regressions before and after changes go live.

• Partner with Product, Operations, and Engineering to prioritize and validate improvements.

• Design lightweight experiments and measurement approaches for model, STT, prompt, workflow, and configuration changes.

• Build visibility into agent performance through dashboards, analytics, and reporting.

• Use AI-assisted development tools and agentic workflows to accelerate experimentation and delivery.

Qualifications

• 5+ years of software engineering experience building production systems.

• Experience shipping Agentic systems into production and measuring whether they worked.

• Strong analytical and problem-solving skills with an ability to translate ambiguous problems into measurable outcomes.

• Ability to use metrics, logs, qualitative review, or experiments to understand whether product changes worked.

• Strong product instincts and ability to connect technical decisions to business outcomes.

• Demonstrated ability to independently own initiatives from problem identification through measurable impact.

• Experience using modern AI-assisted engineering workflows (e.g., Claude, Cursor, Codex, or similar tools).

Nice to Haves

• Experience building eval frameworks for AI products.

• Familiarity with Langfuse, observability platforms, analytics systems, or experimentation tooling.

• Experience with RAG, agent orchestration, prompt management, or workflow automation systems.

• Experience measuring and improving customer-facing AI experiences.

• Healthcare technology or regulated-industry experience.

• Experience working in startup environments with significant ambiguity.

What Success Looks Like

Within your first 90 days, you will:

• Own a product improvement from problem identification through production rollout and measured impact.

• Establish practical checks for evaluating changes before production deployment.

• Create visibility into key Access Agent performance metrics and breaking points.

• Ship measurable improvements with documented before-and-after impact.

• Help Health Note build a disciplined, scalable system for continuous AI improvement.