W.R. Berkley

Their underwriters were retyping records. I debugged the live apps and automated the path between the systems they already had.

W.R. Berkley’s staff live in underwriting web apps that have to match policy, claims, and billing. I shipped features on those apps, fixed production bugs across the TypeScript front end, the GraphQL APIs, and SQL, and removed the manual rekeying between them.

Problem

This was not a greenfield build. The customer’s systems of record already existed, and the apps on top of them were lying by lag: a person copied a field, sometimes wrong. A new portal would have been ignored. The deployment had to land inside the stack they use every day.

Process

I watched the entry. The fields people copied by hand were the integration. I wired the apps to policy, claims, and billing over REST and GraphQL so the screen showed what the core systems held. Production bugs got traced to the layer that was wrong, not patched on the screen that displayed them. QA and DevOps stayed on the Jenkins path they already ran, with Docker and Kubernetes.

Solution

• Integrations into the systems of record, instead of a fourth database.
• Production debugging across the UI, the API, and SQL.
• An automated path for the rekeying, so the error stopped being a person’s job.
• The customer’s existing CI. I did not ask them to adopt a new release ritual.

Results

• 4,500+ Hours a year removed
• 84+ Hours a week
• 3 Systems already in the building: policy, claims, billing
• 0 New system of record

Justin Moran · Forward Deployed Engineer · Chicago, IL · LinkedIn · GitHub · Resume (PDF)