Enterprise CRM Modernization

Legacy WinForms + stored procedures → .NET Web API + React, for a large US logistics company

stack: .NET · C# · React · SQL Server · LINQ · Azure DevOps · LaunchDarkly

──situation

A US logistics company runs its customer operations on a decades-old WinForms application backed by hundreds of undocumented stored procedures. The business logic exists only in production behavior: no specs, no original authors. The mandate: rebuild it as a modern .NET Web API + React platform without regressing a single workflow the business depends on.

──evidence

Exhibit 1: 6+ quarterly releases
sustained delivery cadence with end-to-end ownership: SQL design, API, React UI, release support
Exhibit 2: user-visible timeouts
critical customer screens timed out on large datasets before query and endpoint optimization
Exhibit 3: fragmented service layer
redundant auth layers and base64-encoded payloads accumulated across years of patchwork services

──diagnosis

The risk was never the new stack: it was silent behavioral drift. Legacy stored procedures encoded business rules nobody could enumerate. Every migrated module needed its legacy behavior reverse-engineered, characterized, and proven equivalent before the WinForms version could be retired.

──intervention

01
Reverse-engineered undocumented stored-procedure behavior and rebuilt customer-management modules (approval workflows, classification, billing and payment surfaces) as .NET APIs + React, at feature parity.
02
Led the stored-procedure-to-LINQ migration and consolidated fragmented services into a single experience API, removing redundant auth layers and base64 overhead.
03
Optimized production SQL and endpoints: critical screens went from user-visible timeouts to sub-second responses; resolved recurring cold-start and pagination failures on large datasets.
04
Served as release support lead across quarterly releases: production issue queue, hotfix cherry-picks onto release branches, change-approval coordination.
05
Built custom MCP servers (Python + Playwright) that extract ITSM tickets and generate change-management artifacts, cutting hours of release paperwork per cycle.
06
Defined a 6-phase forensic investigation methodology (data-first, refutation-required) now used to diagnose and document production incidents.

──outcome · verified

Exhibit 1: < 1s
response time on previously timing-out critical screens
Exhibit 2: 0 regressions
legacy workflows replaced at parity across releases
Exhibit 3: hours/release
saved by MCP-automated change-management paperwork
Exhibit 4: stable defect ratio
across releases under mandatory AI-assisted + peer review

Client and internal identifiers withheld under NDA. Scope and outcomes as published in my CV.