Work

Releases that took days, shipped in hours

A lending platform where every deploy was a manual, multi-day event.

Days → HoursDeploy cycle
IndustryFintech
ProductLoan origination and collections SaaS

Anonymised: client names and identifying details are left out. Outcomes are the measured results.

  1. Problem

    • Builds ran on laptops and were not reproducible.
    • Tests were skipped under deadline pressure.
    • Production promotion was manual and risky, so releases bunched up.
  2. What I did

    • Dockerised builds with layer caching so every build is identical and fast.
    • Automated test gates that block a release instead of relying on memory.
    • Zero-touch promotion from staging to production.
  3. Outcome

    • Deployment cycles went from days to hours.
    • Releases became small and routine instead of rare events.
GoDockerCI/CDAWS

More work

Three advisory clients at a time.

Ask about current availability. The first call is free and takes 30 minutes.

Book a Call