Healthcare claims parsing and SQL optimization at Cavista

Context

As a software engineer at Cavista (healthcare), I worked on the systems that parse and process electronic Health Care Claim (837) data, translating Companion Guides into parser code and handling diverse claim segments through EdiWeave and custom models.

Problem and constraint

Claims data is messy and high-stakes: parse it wrong and downstream submission fails. The work had to expand support for more claim segments, keep queries fast under load, and stay reliable across environments, all without regressing existing behavior.

Approach and key decisions

• Optimized SQL written through Entity Framework / LINQ, adding AsNoTracking on read paths and analyzing queries with LINQPad to remove inefficiencies.
• Handled race conditions with techniques like lazy caching and pseudo-random key generation, improving a key process by roughly 80%.
• Built confidence with thorough unit and integration tests (xUnit, NSubstitute, Fluent Assertions) so changes shipped without surprises.
• Closed the loop in production: analyzed logs in Kibana across dev, staging, and production to verify behavior after each release.

Outcome

Broader claim-segment support, faster queries, and reliable releases, with two senior colleagues later vouching for the work by name on LinkedIn.

What I'd do differently

Push more of the race-condition handling into the data model itself rather than application-level guards, so the invariants are harder to bypass.