Brett Buskirk · Platform Engineer
Senior infrastructure work at startup speed — I pair deep engineering judgment with agentic tooling to ship in days what used to take weeks.
Services
Bounded, fixed-scope engagements — each maps to real infrastructure problems I've already solved.
Stand up a production-ready cloud environment from scratch — networking, compute, security, and deployment automation, all codified so it's reproducible from day one.
~2–4 week fixed-scope engagement
Move your app off legacy, on-prem, or basic managed hosting onto a clean, modern, automated cloud foundation — minimal downtime, no mystery infrastructure left behind.
Scoped per project, milestone-billed
Turn click-ops infrastructure into clean, version-controlled Terraform — so changes are reviewable, repeatable, and not one person's tribal knowledge.
~1–3 week fixed-scope engagement
Finally see what production is doing. A complete metrics, logs, and alerting stack — deployed as code — so you catch problems before your users do.
~1–2 week fixed-scope engagement
Ship faster and safer with an automated delivery pipeline — from commit to deploy — on GitHub Actions or GitLab CI.
~1–2 week fixed-scope engagement
Help your team adopt AI coding agents safely and productively — guardrails, CI integration, and the discipline that keeps agents shipping real work instead of driving into a ditch.
Focused 1–2 week engagement
Web apps, PWAs, and internal tools, built to the same standard as the infrastructure underneath them: tested, CI-gated, observable, deployed as code. When the product itself needs AI, it gets built and governed to the same discipline as everything else I ship.
Scoped per project, milestone-billed
Every engagement — agentic work especially — runs on a governance foundation. See how I approach AI risk →
Selected Work
Five projects: the problem, what was built, and the tools that shipped it.
Problem
Generic invoicing tools (Wave, FreshBooks, QuickBooks) don't match a solo technical consultant's actual pipeline: proposal → SOW → deposit → milestones → retainer → tax set-aside, with contracts and client context scattered across apps. The value is niche fit, not more features.
What was built
A local-first PWA that runs the entire back office in one connected data model where everything links to a client. React 19 + TypeScript with Dexie (IndexedDB) for reactive offline data, installable and fully offline-capable, with PDF invoice/document generation, opt-in client-side encryption, encrypted backups, and a command-center dashboard. No server, no accounts — the data lives on your machine.
Problem
Build an offline-capable, phone-first reentry simulation for returning citizens — complex branching logic, 70 life events, and 161 automated tests, deployed to production with zero manual steps.
What was built
React 19 / TypeScript / Vite PWA with Dexie for offline state and serializable run codes. GitHub Actions CI runs typecheck, tests, and production build on every PR. Merge to main: DigitalOcean App Platform picks it up automatically via .do/app.yaml.
Problem
A publishing site on WordPress + Elementor was expensive to maintain, impossible to customize without fighting the platform, and returning poor Lighthouse scores.
What was built
Custom Astro 5 static site with Tailwind CSS v4, Pagefind full-text search, and full content migration of 22 articles with optimized images. GitOps deployment: push to main, DigitalOcean App Platform rebuilds and ships automatically.
Problem
Most small teams run services across multiple droplets with zero centralized visibility: no metrics, no log aggregation, no alerts, until something breaks and there's no way to see why.
What was built
A complete observability stack provisioned entirely as code: Terraform stands up a DigitalOcean VPC and management node, Ansible installs Prometheus, Grafana, Loki, and Alertmanager via Docker Compose, and every node ships metrics and logs back privately. Reachable only over Tailscale: zero public management surface. Open source and drop-in: one terraform apply and one ansible-playbook run away from a working observability hub for any team.
Problem
AI coding agents open pull requests fast — but they leak secrets, drift out of scope, skip tests, and slip in dependencies. Teams either rubber-stamp the work or burn senior time reviewing every line.
What was built
A CI guardrail — GitHub Action plus CLI — that inspects every agent PR's diff for risk signals, posts a structured review comment, and sets a pass/fail check. Six deterministic rules plus an opt-in LLM intent check (AI reviewing AI). Published to npm and the GitHub Marketplace, with 99%+ test coverage enforced in CI.
These five are the highlight reel — there's more where they came from. Check out everything →
About
I'm a platform/DevOps engineer who builds and runs production systems end to end: the application, the cloud infrastructure under it, and the pipeline that ships it. I help startups and small teams get all three (Terraform, Ansible, CI/CD, observability, and the application layer itself) on a project basis, so they get one senior engineer instead of hiring an app contractor and a platform person separately.
Five years building production systems across AWS, GCP, and DigitalOcean, with a full MERN-stack background behind the application side of that work. I lean on AI coding agents as horsepower under that judgment — which means startup-grade speed without giving up the operational rigor production demands.
Stack
Contact
Prefer to talk it through? Book a 30-minute call — free, no prep needed, bring the problem you're solving. Or send a message below and I'll get back to you within one business day.
Thanks for reaching out. I'll get back to you within one business day.