Production platform modernization for a growing SaaS
A multi-service production stack rebuilt for reliability: zero-downtime deploys, automated infrastructure, and workflow automation the team can operate themselves.
Representative engagement. This story describes work we do, anonymized and with details changed. We publish named case studies only with client approval.
Context
A growing SaaS company had the classic scale-up inheritance: several services deployed by hand, SSL certificates renewed by calendar reminder, and a deploy process that required the one engineer who knew the incantations. Every release was a small act of courage.
The challenge
The platform couldn't be taken offline to be fixed — customers were on it. The modernization had to happen alongside normal feature delivery, service by service, with each migration reversible. And the end state had to be operable by the company's own small team, not dependent on us.
What we built
- A containerized production stack — every service in Docker with reverse-proxy routing, automated SSL, and per-service health checks replacing the hand-maintained server setup.
- Zero-downtime deployments — rolling deploys with automatic rollback on failed health checks, turning releases from an event into a non-event.
- Workflow automation — internal operations (onboarding, alerts, data syncs) moved from manual runbooks into automated n8n workflows the team can edit visually.
- Documentation as part of delivery — architecture notes, runbooks, and incident playbooks written as we went, so the handoff was a formality rather than a project.
How it went
The migration completed service by service with no customer-facing downtime. Deploys went from a scheduled, supervised ritual to something any engineer on the team runs multiple times a day. Six months later the stack is operated entirely in-house — which was the point.
Why it's representative
Platform work is judged by what stops happening: no more deploy anxiety, no more certificate surprises, no more single point of human failure. This engagement is representative of how we approach infrastructure — incremental, reversible, and designed for the client's team to own.