An AI agent for coverage validation in CI/CD

A pipeline agent that reads deployment descriptions and repository metadata to recommend missing tests.

Organisation
JPMorgan Chase
My role
Designer and engineering leader
Period
Jul 2024 – Nov 2026

The problem

Before a deployment, someone has to check whether the change is properly covered by tests. Done manually for every deployment, that review is time-consuming and depends on the reviewer knowing both the change and the test suite well.

My role

I designed the agent as part of the AI platform and agentic architecture work for the platform team.

Approach and architecture

The agent runs as part of CI/CD. It parses the deployment description and the repository metadata for a change and recommends the tests that are missing, so engineers start their coverage review from a targeted list instead of from scratch.

Evaluation and governance

The agent recommends tests; engineers decide what to add. That keeps a person accountable for coverage, consistent with the human-review controls embedded in the GenAI workflows I led. See LLM evaluation and responsible AI controls.

  1. Inputs
    • Deployment description
    • Repository metadata
  2. Analyse
    • Coverage agentRuns in CI/CD
  3. Recommend
    • Missing testsTargeted recommendations
  4. Decide
    • Engineer reviewChooses what to add
Illustrative CI/CD coverage validation flow. A simplified reconstruction from the resume, not a depiction of any confidential system.

Documented outcomes

  • Cut manual coverage review by 40%.
  • Agentic AI
  • AI Quality
  • Engineering Leadership

Harness Engineering - Beyond the AI Model

AI can identify a failure and suggest a fix. Should we trust it to fix the problem on its own? Why AI-driven remediation needs production-grade discipline.

· 1 min read