I build AI-flavored products end-to-end — from pixel to prod.
Email · Website · LinkedIn · GitHub
I ship fast, keep the UI calm, and let the architecture carry its weight.
Most days I’m toggling between React/SvelteKit, Node/Python, and a sprinkle of cloud/devops to get ideas into users’ hands—quickly.
Core: TypeScript, React, Svelte/SvelteKit, Node.js, Python
Also comfortable with: Next.js, Vue/Nuxt, Django, FastAPI, Flask
Data & infra: PostgreSQL, SQL/NoSQL, Redis
Cloudy stuff: AWS, Docker, Vercel, Netlify, Upstash, CI/CD, GraphQL, REST, tRPC
I pick tools for the problem, not the other way around.
- Useful AI features that assist, not overwhelm.
- Small services that talk well to each other (serverless when it helps, boring monoliths when it’s wiser).
- Interfaces that feel obvious on first click.
- AI SaaS for high-stakes analysis — handled data, infra, and the tiny UI details that keep it approachable.
- Health app features used by real people every day — serverless CRUD, helpful docs, fewer support pings.
- Presentation/ops tools that make teammates look good in demos and ship rooms.
(If you’re curious about architecture choices, I wrote about SvelteKit + tRPC for scalable apps → https://michaelbelete.com/sveltekit-trpc-scalable-apps/)
- Ship the smallest useful thing. Measure. Then ship again.
- Make UI quiet. Remove decisions users shouldn’t make.
- Logs are a UX too. DevEx is part of product.
- Choose boring when it’s safer. New ≠ better.
I’m into: AI-assisted products, clean UI, and systems that won’t wake you at 3 AM.
Say hi: it.michael.belete@gmail.com