Highlights
- Automated high-volume business-rule validation against contractual and business requirements in PY25, offsetting ~30 FTEs of manual effort while validating $40M in billable events
Project Narrative
On the VA MDE project, once a disability benefits questionnaire (DBQ) passed QA, the associated contention entered Final Contention Review (FCR): the stage where a billing team confirmed the contention met all contractual requirements before it was sent on to the VBA. That review meant verifying the necessary DBQs were complete and completed correctly, that the CLIN codes were correct for invoicing, and that required business logic, diagnostic checks and form completeness, had been satisfied. Because the account runs on transactional pricing, FCR is the gate between completed work and getting paid for it. Every contention that clears it is a billable event, so the review had to be both correct and fast, and it scaled directly with the account’s volume.
I designed a backend to automate the FCR compliance checks and led the development team to implement. Because the work is essentially high-volume business-rule validation followed by an API call, and because contention volume on the account is very large with difficult burst periods, I wrote it in Rust rather than Python, leaning on async throughout to maximize throughput and absorb load spikes efficiently. The engine validated contention data held in MSSQL and Postgres against the contractual rule set before submission. It deployed initially on AWS Lambda (with images in ECR), and when volume outgrew that model (15 min runtime limitation), I moved it to AWS ECS. Because FCR requirements shift as the finance team’s contractual rules change, the system is built to be updated continuously against their evolving specifications.
The design principle was deliberate: automate the routine, high-volume checks so the billing team could concentrate on the genuinely hard cases. In PY25 the system checked $40M in billable events against their business and contract requirements before invoicing, automating an estimated ~30 FTEs of review effort, and formed part of the automation portfolio that drove ~$35M in margin on $170M+ in revenue.