Full-Stack AI Engineer building production-minded LLM apps
I build AI products that go beyond the demo: structured outputs, validation, evals, guardrails, full-stack SaaS, automation systems, and production-focused architecture.
A live LLM lead-analysis app that turns a company domain into a structured lead card with evidence, confidence notes, and follow-up angles.
Built with structured outputs, Zod validation, repair retries, degraded states, SSRF protection, prompt-injection boundaries, and production metrics.
TypeScript Next.js Zod LLM APIs Vercel
A prompt-evaluation and LLM API system for scoring, improving, and routing prompts through structured quality gates.
Python FastAPI Next.js Vertex AI Gemini Cloud SQL
A full-stack SaaS platform for B2B research workflows, job orchestration, validation, auth, billing, and real-time monitoring.
TypeScript Next.js PostgreSQL Prisma BullMQ Stripe Clerk
A Manifest V3 Chrome Extension for keeping structured character profiles consistent across AI generation tools.
TypeScript Manifest V3 Vitest Playwright
A live multilingual marketplace connecting Georgian local producers with buyers.
Next.js Supabase PostgreSQL next-intl Tailwind Vercel
I use AI-assisted engineering workflows, but I keep the judgment human:
- define the product problem before implementation
- write architecture and acceptance criteria before code
- use AI agents for implementation support
- validate outputs with tests, reviews, and manual QA
- keep production risks visible: auth, data shape, failure states, security, observability
- LLM apps with structured outputs and validation
- RAG/eval reliability and production debugging
- Full-stack SaaS with Next.js, FastAPI, PostgreSQL, Supabase, Prisma
- Auth, billing, dashboards, APIs, workers, and deployment
- Browser automation and workflow tools
TypeScript · Next.js · React · Python · FastAPI · PostgreSQL · Supabase · Prisma · BullMQ · Docker · Vercel · LLM APIs