Not a list of buzzwords. A toolkit that ships.

Anyone can paste a list of technologies onto a page. What matters is whether those technologies have shipped anything that mattered under real deadlines, real users, and real money on the line. Everything below has.

This page isn't organized the way a job board would sort it. It's organized the way I actually work — by the kind of problem each skill set solves.

AI & Intelligent Systems

For when "search" and "support" need to actually understand what someone is asking.

  • RAG Architecture — Naive, Conversational, and Agentic retrieval-augmented generation, built to retrieve the right answer, not just a plausible one
  • pgvector — semantic search at production scale, embedded directly in PostgreSQL
  • OpenAI — custom tools, function calling, and multi-step agent design
  • Gemini Live — real-time, low-latency voice agents
  • LangChain — orchestration for multi-step LLM workflows
  • Ollama — local/self-hosted model deployment where data can't leave the building

Where this shows up: an Agentic RAG pipeline that cut clinical protocol lookup time by 60% across 100,000+ patients — proof this isn't theoretical.

Automation & Workflow Engineering

For when growth is bottlenecked by manual work, not by demand.

  • n8n — Verified Creator status; workflows with 60+ nodes running in production, not demo pipelines that fall apart at scale
  • AI-powered qualification pipelines — routing, scoring, and enriching leads before a human ever touches them
  • Third-party orchestration — Stripe, Twilio, Vapi, Google OAuth, Mapbox wired together into a single, reliable workflow

Where this shows up: a 68-node automation pipeline that tripled qualified B2B lead throughput for a client in Oman — without adding headcount.

Frontend Engineering

For when the product needs to feel instant, even when the data isn't.

  • React & Next.js — SSR, ISR, and component architecture built for scale, not just for the demo
  • TypeScript — strict-mode by default; type errors caught before they cost you a customer
  • TailwindCSS, shadcn/ui, Radix UI, Framer Motion — interfaces that feel considered, not templated
  • Redux Toolkit, TanStack Query, Zod — state and data-fetching that don't quietly rot as the app grows
  • Vite — fast builds, faster iteration

Where this shows up: a 70% render-performance improvement on 10,000-row data grids, and an 85% drop in production type errors after enforcing strict TypeScript across an entire codebase.

Backend & Infrastructure

For when the product needs to survive its own success.

  • FastAPI, ElysiaJS, ExpressJS, Bun, tRPC — APIs built for speed of iteration and reliability under load
  • PostgreSQL, Prisma, Drizzle ORM, Redis, MySQL, Neon — data layers that scale without turning into a migration nightmare
  • AWS (EC2, S3, RDS), Nginx, Docker, GitHub Actions, CI/CD — infrastructure that deploys in minutes, not hours, with zero downtime

Where this shows up: a CI/CD pipeline that cut deploy time from 2 hours to under 12 minutes with zero-downtime blue-green deploys, and a multi-tenant system holding 99.9% alert delivery reliability.

Product & Design Tooling

For when engineering and design need to actually speak the same language.

  • Figma — translating design intent into production-accurate builds
  • Cursor AI — AI-assisted development without sacrificing code quality
  • Biome, ESLint, Husky — code quality enforced automatically, not by hoping people remember

The Honest Summary

If your business needs an AI layer that actually understands your data, an automation system that replaces hours of manual work, or a product that doesn't buckle the day it gets popular — this is the toolkit that gets it there. The tools change every year. The standard for what "done" means doesn't.

Curious how these come together on an actual project? See selected work, or get in touch to talk through what you're trying to solve.