Amtarc Logo
0%
Home/Case Studies

Self-Service Deployments for a 40-Person Engineering Team

How deployment bottlenecks were eliminated by building a lightweight IDP with golden paths — increasing deploy frequency by 12x.

Platform Engineering IDP
12x
Deploy Frequency
80%
Lead Time Reduction
40
Engineers Served
< 15 min
New Service Bootstrap
Context

The Problem

A scale-up SaaS company had grown its engineering organization from 10 to 40 developers in 18 months. Their deployment process hadn't scaled to match. Deploying a single service required developers to file tickets with the platform team, navigate multiple manual steps, and wait days for their changes to reach production.

Developer satisfaction scores were declining. The platform team, overwhelmed by operational requests, had no bandwidth for strategic infrastructure work.

Key Constraints

1

Platform must support their evolving microservices footprint

2

Must not require developer knowledge of Kubernetes or Terraform

3

Existing CI/CD pipelines (GitHub Actions) must be preserved

4

New service bootstrapping must take less than 15 minutes end-to-end

Architecture

The Solution

The solution centered on a Backstage-powered Internal Developer Platform with opinionated golden paths for the company's core service archetypes. Developers never interact with Kubernetes directly — they interact with a curated catalog of software templates.

Deployments are orchestrated by ArgoCD with GitOps-driven configuration. Crossplane provisions cloud resources (databases, queues) through the same Kubernetes API, giving developers self-service infrastructure without cloud console access.

Technical Architecture

layers

Backstage Developer Portal

Centralized portal with Software Catalog, TechDocs, and templates. Single pane of glass for all engineering resources.

rocket_launch

ArgoCD GitOps Engine

All deployments declared in Git. ArgoCD continuously reconciles desired vs actual state with automatic drift correction.

cloud_queue

Crossplane Infrastructure

Platform team provides databases and queues as Kubernetes CRDs. Developers provision via YAML.

policy

Policy-as-Code

OPA Gatekeeper enforces security, resource limits, and naming conventions at the Kubernetes API level.

"The team used to dread shipping features. Now they ship daily with confidence. The platform fundamentally changed the engineering culture."