Fortune-10 tech company · 2026
Found the root cause of two production outages in a global internal operations platform — one query run 340,000 times — and drove the resilience plan.
- Client
- Fortune-10 tech company
- Role
- Database & performance specialist
- Year
- 2026
Outcomes
Context
A business-critical internal platform went down for about nine hours, then again three weeks later. I led the database investigation and wrote both postmortems.
What I built
- 01
Root cause: a single user-role query executed 340,187 times at ~2.9 s each — ~278 hours of database time — over a dataset small enough to fit in RAM thirty times over.
- 02
PostgreSQL index and configuration recommendations, and review of the circuit-breaker and cache implementation.
- 03
Resilience plan, implemented by the team: database and client-side circuit breakers, connection-pool hardening and a user/role cache.