Product Engineer

Role Summary

We’re seeking a Product Engineer who works the way software engineering is heading: directing AI coding agents to produce the work, then applying the judgment, quality bar, and domain knowledge to make sure what ships is correct. You will be responsible for the research fulfillment systems and the data analysis interfaces that sit on top of them — front-end surfaces used across the data analysis workflow, the backend pipeline that powers them, and the agent supervision layer that governs AI execution throughout the fulfillment lifecycle.

The difference between this role and a traditional IC is how the work gets done: you use AI coding agents as your primary development tool, you direct what they build, you review what they produce, and you own the quality of what goes to production. The technical depth this requires is real — you cannot supervise an agent writing fulfillment pipeline logic if you do not know what correct fulfillment pipeline logic looks like.

• Use AI coding agents as your primary tool — writing precise specs, reviewing all output before it ships, and overriding when code is functionally correct but architecturally wrong.
• Decompose complex requirements into well‑scoped tasks, knowing what to delegate versus handle directly.
• Hold agent‑generated code to the same standard as handwritten code.

• Own the pipeline from intake through delivery — order configuration and routing, sample allocation, quota management, study execution, fielding logic, and data handoff.
• Specify and validate the business logic agents implement (quota balancing, incidence rate fallbacks, fielding quality controls, data contracts), catching cases where output meets the literal spec but misses operational intent.

• Direct agent development of the React‑based interfaces for study configuration, execution monitoring, and results access — reviewing for correctness, usability, and design consistency.
• Specify human‑in‑the‑loop interfaces where operators approve AI‑initiated actions, with enough clarity for agents to build and enough judgment to catch ambiguity or risk.

• Define and maintain the guardrails governing what agents may do autonomously.
• Own the observability and failure recovery systems — logging, alerting, and recovery — that keep agent behavior visible, diagnosable, and recoverable without manual heroics.

• Take personal accountability for everything that ships.
• Triage and remediate technical debt using agents to execute the work.
• Develop the prompting patterns, specification templates, and review checklists that make agent‑directed work faster and more reliable.
• Own incident response so each class of failure is permanently retired.

• Work with product and operations to turn requirements into agent‑ready specifications.
• Coordinate with data pipeline and platform teams on interface contracts and data formats.
• Feed production patterns back into both the roadmap and the development process.

• 3–6 years of software engineering experience with genuine full‑stack depth — you have owned systems across front‑end and backend, understand how the pieces connect, and can recognize incorrect implementations at code review speed across both layers.
• Hands‑on experience working with AI coding agents — Claude, Cursor, Copilot, or similar — as a primary development tool. You know how to write specifications that produce reliable agent output, how to identify where agents go wrong, and how to course‑correct efficiently without starting over.
• Sufficient React and Type Script depth to review agent‑generated front‑end code critically — catching component architecture problems, state management errors, and performance issues that agents commonly introduce when given under specified requirements.
• Sufficient backend depth — API design, data modeling, workflow state management, job orchestration — to specify fulfillment pipeline logic precisely enough for agents to implement correctly, and to catch when agent output…