Lab

LAB

Side builds, experiments and open source.

$ /project-init$ /code-review$ /deploy-checklist$ /client-report

FREELANCER DEV TOOLKIT

I was repeating the same setup steps on every new project - scaffolding, linting, deploy scripts, client reports. I productized my workflow into a reusable toolkit: custom Claude Code commands for project init, automated code review, deploy checklists, and client-facing reports. Saved 2+ hours per project start.

Claude CodePythonShellGit

▶ Watch Demo GitHub

Tools

STACK

ai_llms.sh

→Claude API + Claude Code (daily driver)
→OpenAI API (GPT-5.5, Vision)
→Gemini 3.1 Pro (Vertex AI, video understanding)
→Multi-agent orchestration (LangGraph, CrewAI, MCP)
→Voice & realtime: Speechmatics RT/batch, Whisper, ElevenLabs
→RAG: FAISS, Supabase pgvector, Pinecone, hybrid search + re-ranking
→LLM Evals (DeepEval, LLM-as-Judge, grounding checks)
→Prompt engineering & adversarial agent design

automation_data.sh

→n8n (self-hosted) / Make.com (Advanced Certified)
→Python · FastAPI · SQLModel
→PostgreSQL / Supabase · Airtable
→REST APIs · Webhooks · MCP servers
→Docker · Vultr · GCP · Vercel
→Git · GitHub Actions (CI/CD)
→PostHog · Sentry · structured logging & monitoring
→Next.js · TypeScript · Tailwind · shadcn/ui

product_delivery.sh

→PRINCE2 / Agile / Scrum
→MVP Scoping & Roadmapping
→Linear / Jira / Notion
→Figma / FigJam / Miro
→User Research (Maze, Typeform)
→A/B Testing & Experimentation
→Perplexity Pro (Competitive Research)
→Loom (Async Communication)

ai_llms.sh

automation_data.sh

product_delivery.sh