Peak-season releases without the fire drill
A commerce brand tightened CI/CD, observability, and squad practices ahead of their busiest quarter.
The challenge
Promotions drove traffic spikes that exposed fragile deployment scripts and blind spots in monitoring. The team needed confidence before Black Friday, not another heroics weekend.
How we approached it
Prioritise release risk
Mapped release paths and incident history to prioritise the riskiest manual steps.
Pipeline automation
Automated build, test, and promotion pipelines with feature flags for safer launches.
Observable checkout paths
Added synthetic checks and SLO dashboards tied to checkout and catalogue services.
Sustainable handover
Coached internal engineers on runbooks so improvements stuck after Agam rolled off.
Outcomes
Faster releases
Release cadence improved within the first sprint cycle.
Clean peak weekend
Peak traffic weekend completed without Sev-1 incidents for the first time in three years.
Quieter on-call
On-call noise dropped after alert tuning and ownership boundaries were clarified.
Technologies involved
- Next.js
- Node.js
- Redis
- AWS
- Datadog
- GitHub Actions
In their words
“Our dedicated squad improved release cadence within the first sprint cycle.”