CW DevOps — Internal Developer Platform

What it is

The internal developer platform that runs my globally distributed team (Ghana, Rwanda, US) — and my engineering methodology, productized. It standardizes how software gets built, shipped, and governed, regardless of location or a client's toolchain. ~32,000 LoC, 2,682 tests.

Provider-neutral by design

38 provider implementations across 11 categories — project management, VCS, CI, artifact, deployment, notification, docs, test-management, infrastructure, LLM, and coding — swappable at runtime via Python protocols. JIRA, GitHub, GitLab, Azure, Bitbucket, Gitea, Linear, Jenkins, CircleCI, Nexus, and more. No platform lock-in.

Self-service infrastructure (Terraform)

Bundled Terraform for AWS, Azure, and GCP — product templates (static-site, API service, web app) and reusable modules (network, database, cache, storage, compute, monitoring) provisioning VPC, RDS, ECS, ALB, IAM, ElastiCache, and CloudWatch. Engineers scaffold golden-path infrastructure without writing Terraform by hand.

One pipeline, six CI platforms

A Pipeline IR (intermediate representation) renders a single definition to GitHub Actions, GitLab CI, Azure Pipelines, Bitbucket, Jenkins (Groovy), and CircleCI — so teams change CI providers without rewriting pipelines.

Governed AI delivery

An autonomous Claude worker runs the full loop — ticket → branch → implement → tests → auto-retry on failure → merge request → optional auto-merge — but a compliance engine blocks the merge on any violation (commit-status gate, violation tracking with badges). The AI moves fast inside controls it cannot bypass. Exposed to AI clients via an MCP server (~30 tools).

Meets developers where they work

Four developer surfaces, one source of truth:

• CLI — 34+ commands across the full lifecycle.
• VSCode extension — an Activity Bar sidebar (tickets, pipelines, health, analytics, infrastructure) with an embedded AI assistant. CLI-driven, so it inherits the same providers, auth, and config — no separate setup.
• Web UI — operations dashboard.
• MCP server — ~30 tools, so external AI agents drive the platform directly.

Auth and configuration live in one place; the terminal, the editor, the browser, and the agents all read the same state.

Governance & insight

Policy-driven quality gates (YAML + JSON-Schema), DORA metrics (deploy frequency, lead time, change-failure rate, MTTR) and AI-productivity analytics, and multi-tenant isolation (per-tenant data, quotas, plans). Self-hostable and air-gappable — bundled Docker Compose stacks for GitLab, Gitea, Jenkins, and Atlassian.

The wedge: no single product unifies a provider-neutral developer platform, multi-CI pipeline IR, Terraform self-service, and governed AI autonomy — self-hostable for regulated environments.