Fabrizio Álvarez
← All case studies

Performance

Performance & reliability under load

Cut a critical database query from 30s to ~2s, hardened a corrupted-cache path, and ran 100+ production hotfixes with disciplined, reversible changes.

Sanitized excerpt of commercial work on Salutte2 (Alephoo HDE) — an EMR SaaS where some clinics process up to 30,000 appointments/day. No proprietary code is disclosed.

Slow query optimization: 30s to ~2s

Context

At scale, a few slow queries and a couple of corrupt-cache incidents degraded the experience in the highest-volume clinics. Reliability is critical in healthcare software — a degraded appointment flow blocks patients.

What I did

  • Cut critical database queries from >30s to ~2s (~15×) by rewriting them and adding the right indexes — the single most-quoted metric from this work.
  • Diagnosed and fixed corrupted-cache incidents and hardened the cache path.
  • 100+ production deploys and hotfixes, including urgent incident response.
  • Consolidated error monitoring on Sentry so regressions surface immediately rather than via user reports.

Decisions worth defending

  • Measure, then optimize. The 30s→2s win came from identifying the actual offending queries (via slow-query analysis), not speculative indexing.
  • Treat production access as a first-class skill — disciplined, small, reversible hotfixes under pressure, not heroics.

Outcome

Materially better UX in high-volume clinics and faster incident detection and resolution across the platform.

Evidence (public PR metadata)

  • PR #13795 — Production mejoras slow queries.
  • PR #13468 — corrupted-cache fix.
  • 100+ PRs tagged Production.