I started in a full-time non-dev job, taught myself to code, became a frontend dev, now full-stack in the AI space at one of the largest software companies in the world. I ship side projects in public under the Hackastak umbrella: CLIs, AI agent tooling, workflow automations, and digital products on Gumroad.
Side projects under the Hackastak umbrella. Built in public, shipped on weekends, and paid for in coffee. Agent tooling and CLIs you can install today.
The agent harness I use every day, packaged so anyone else can: 155 skills, 21 agents, 53 slash commands. The hard part isn't the content, it's portability — one source of truth that installs cleanly into Claude Code, Cursor, Codex, and OpenCode, which all disagree on config format.
Semantic search and RAG across every repo you own or starred, without shipping your code to anyone. It all runs locally: a statically-linked Go binary embeds SQLite plus the sqlite-vec extension through CGO, and pointing embeddings at Ollama means zero network calls at all.
A subscription SaaS end to end, not just invoice CRUD: Stripe Checkout for plans, a webhook route that reconciles subscription state server-side, Supabase SSR auth, and Drizzle-typed queries. Invoice PDFs are generated with jsPDF rather than by spinning up a headless browser.
Honest, no-BS recommendations on the tools and practices worth your time. Developer tooling, AI x dev workflow, engineering best practices, and career notes from a self-taught dev still in the trenches.
I use Claude Code and Cursor every day, and I still edit in Neovim. That combination confuses people, so here’s the honest accounting of why Vim earned its spot in an AI-first…
My Obsidian vault had become very good at proving me right. I could search almost any opinion and find years of notes ready to support it. That felt useful until I realized what…
The fastest engineering lessons in my week usually start in someone else’s code Image generated by AI Most Mondays, I open GitHub and find a line of pull requests waiting before…
Part 3 of Demystifying Prompt Engineering Part 1 gave us the mental model. The LLM finishes documents. Part 2 used it to write a better single prompt. This part is about everything…
From a non-dev full-time job to full-stack engineering at a major AI software company — the timeline of the pivot, plus the side-project habit that came along for the ride.
Building production AI agent tooling, MCP integrations, and Claude SDK-powered workflows for enterprise teams. Day-job lessons feed The HackaStak's AI x developer pillar.
Owned the customer-facing application — design system, accessibility, and performance work. Mentored teammates and shipped the features that moved the metrics.
Lead frontend engineer on a React app running on Kubernetes against multiple microservices, used by surgical imaging teams across 30,000+ cases a year. Mentored engineers and senior interns. This is where the pivot landed: the nights-and-weekends learning became the day job.
Led cross-functional product launches for medical devices that grew annual revenue by millions — mechanical engineering, not software. Taught myself to code on nights and weekends here, which is where the career change started.
BS in Mechanical Engineering, then software on nights and weekends. Engineering fundamentals from the degree, the stack from side projects, OSS, and shipping in public. Bootcamp grads, juniors, and career-changers are exactly who The HackaStak is for.
React, Next.js, TypeScript, Tailwind
Node.js, Python, PostgreSQL
Agent reliability, evals, observability, MCP, Claude SDK
Docker, Kubernetes, CI/CD, Vercel