Product Specialist

PRODUCT SPECIALIST
Agent-ready specifications and end-to-end quality ownership in a three-person AI-native delivery squad
Role
Product Specialist
Function
AI-Native Delivery Squad — Product & Quality
Experience
8–12 years, including 3+ years in a combined specification and quality-ownership role
Location
[Location] — hybrid, with overlap into client business hours
Engagement
Full-time · small-squad delivery on a fixed-timeline platform build
Reports to
Delivery Lead / Architect-Orchestrator
About the role
We deliver platform builds with small AI-native squads. Instead of a large team writing production code by hand, a three-person squad specifies the work, programs and supervises AI coding agents, and verifies everything those agents produce. The agents are the workforce; the squad orchestrates and assures.
The Product Specialist owns what gets built and whether it is right. You turn business and partner requirements into specifications precise enough for an agent to implement without ambiguity, then accept the resulting output against intent before the squad ships it. You also own functional, integration and end-to-end quality for the platform.
There is no junior tier and no layer of people writing tickets beneath you. Every specification you write is executed at speed, and every gap in it surfaces within hours. This role suits someone who writes with precision, is comfortable saying an increment is not acceptable, and wants to work at the front of AI-assisted delivery rather than read about it.
Key responsibilities
Specification ownership
Convert business, partner and regulatory requirements into agent-ready specifications — user need, acceptance criteria, API contracts, state transitions, data models, edge cases and negative paths — with no ambiguity left for an agent to resolve on its own.
Own the baseline product specification, the journey and state-machine models, and the acceptance-criteria specification set that serve as executable input to AI coding assistants.
Build and maintain a spec-driven backlog, groomed continuously to stay ahead of agent throughput on a compressed delivery timeline (single-week discovery, multi-week AI-assisted build).
Own requirement traceability end to end — requirement to specification to build to test to sign-off — and produce the traceability evidence pack used at each quality gate.
Specify partner integration contracts alongside the architect: request and response schemas, webhook payloads, callback semantics, error and retry behavior, idempotency rules and correlation identifiers.
Accepting and governing agent output
Accept AI-generated output against intent before the squad ships. Decide per increment whether output is fit to ship, needs rework, or must be rejected.
Diagnose specification weakness when agents drift — tighten the specification rather than patch the output.
Work inside defined agent autonomy boundaries, escalating anything that touches personal or sensitive data, consent wording, authentication and session handling, or third-party partner integrations to human decision.
Contribute to the reusable specification, prompt and playbook library so that patterns proven on one journey are reused across the next.
Functional, integration and end-to-end quality
Own functional, integration and end-to-end test coverage for the platform — this is a specification-and-quality role, not a pure product role.
Design and execute business-rules validation, partner contract and integration testing (schema validation, webhook signature handling, retries, idempotency, correlation IDs), end-to-end journey testing and negative-path coverage.
Apply AI-enabled test tooling across the lifecycle: requirement intelligence and gap detection, AI test-case generation, AI-powered API automation, and browser automation via Playwright or an equivalent MCP-driven tool.