SPEC Engineer

The Institutes•Malvern, PA

2d•Hybrid

About The Position

The Spec Engineer serves as the critical translation layer between business stakeholder intent and machine-executable instruction, ensuring autonomous AI agents receive complete, unambiguous specifications before any code generation begins.

Requirements

3+ years in business analysis, systems analysis, or technical requirements engineering, preferably in an enterprise or regulated environment.
Demonstrated ability to write clear, structured technical documentation (PRD, SRS, or equivalent) with a strong grasp of intent-level versus implementation-level requirements.
Working knowledge of data modeling, API constructs, and business-rule documentation conventions.
Comfort operating within a version-controlled repository (GitHub or equivalent) and treating documentation as code.
Familiarity with AWS or comparable cloud platforms and containerization models such as Docker, AWS ECS, or Kubernetes.
Strong stakeholder-facing communication skills, including the ability to press executives and subject-matter experts for decisions on edge cases without stalling delivery.
Prior exposure to spec-driven development workflows or tools such as Claude Code, AWS Kiro, or GitHub SpecKit.

Nice To Haves

Working familiarity with Python and TypeScript, with comfort in JavaScript and React-based or Express.js/Next.js web development contexts.
Background in software design or architecture sufficient to anticipate implementation ambiguity before it becomes a defect.
Experience with AI-agent tooling including GitHub Copilot, Claude/Claude Code, and markdown-based specification file authoring.
Familiarity with property and casualty insurance terminology, e-commerce workflows, or professional licensure and compliance data models.

Responsibilities

Interrogate ambiguity before implementation by surfacing unresolved edge cases, behavioral choices, rate limits, and failure modes back to business stakeholders.
Write goal-oriented specifications combining PRD-style intent statements with SRS-style precision to ensure agents generate correct, testable code.
Maintain each specification as a version-controlled living artifact, updated continuously as decisions are made or discovered throughout the build cycle.
Own the plan-review checkpoint, ensuring no code generation proceeds until a human has reviewed, edited, and approved the agent's plan file.
Curate a canonical standards library covering data models, business-rules catalogs, and coding standards so every spec references shared definitions rather than re-describing them.
Partner with Architects and Platform Engineers during plan-approval handoffs to ensure approved specs decompose cleanly into executable task graphs.
Facilitate structured requirements sessions with business stakeholders, converting conversational intent into unambiguous, versioned specification language.