cat trilogy.md
Trilogy
Enterprise software group (ESW Capital). Austin, TX.
Trilogy acquires enterprise software products and keeps them running. I am the product expert and engineering owner for six of them: a customer advocacy platform, an employee social advocacy platform, a PaaS hosting provider, two CRM platforms and a customer support suite.
Because they arrived through acquisition, the code predates everyone currently working on it and the documentation arrived in whatever state the previous owner left it. The job is to own that: reproduce what a customer reports, trace it to where the fix actually belongs, ship it with a regression test, and keep the build and release machinery underneath it working.
Since 2025 a growing share of the work is done by agents I built the infrastructure for, which is the part I find most interesting.
- 580
- of 591 defects closed
- 266
- pull requests merged
- 173
- reviewed for others
- 56
- repositories touched
Case studies
- Automating the maintenance pipeline
Built the agent infrastructure that now triages, fixes, tests and reviews defect work across six enterprise products. My own throughput went up fourfold in two years.
- Rebuilding two enterprise build systems
Cut one enterprise CRM build from over three hours to under ninety minutes, then authored all seventeen pipelines for a second platform from scratch.
- A customer-facing product in sixteen days
Empty repository to signed public release in sixteen days, built alone: scans a customer estate, classifies upgrade blockers, computes a viable path and produces a readiness report.
- Owning an iOS release train
Seventeen TestFlight builds in seven weeks for an enterprise iPad client, plus customer-branded packages, and a pre-build check that catches what Apple would reject days later.
- Security work across multi-tenant platforms
Per-tenant CSP enforcement, removing hardcoded credentials from a hosting platform, finally getting 2FA enforced, and hardening the CI supply chain.
- Fixes on hard legacy surfaces
A native C++ crash in a record merge, Exchange connectors duplicating records, spreadsheet export corruption, and Scala polling workers rewritten for a new database driver.
Also did
- Primary responder for production incidents across every assigned product, with production access and root cause analysis ownership.
- Automated customer patch delivery, so validated fixes are built, signed and packaged by the pipeline instead of by hand.
- Made self-hosted build infrastructure survive real failure: file-lock contention, scoped process cleanup, and deliberate fault injection to prove resilience rather than assume it.
- Authored the per-product context layer agents load before touching a codebase, covering all six products, plus the bulk migration tool that moves legacy documentation in with AI-proposed structure.
- Every change traceable to a tracked ticket and paired with a regression test that was observed failing first.