Full-stack AI engineer. I build the app around the model.
Agents, retrieval, browser automation, APIs, and the interface people actually use. I like taking a messy workflow and turning it into something someone can run without me sitting next to them.
Portfolio & demos · LinkedIn · Say hello
BusinessFrame — AI-assisted Google Ads
Website → business analysis → campaign setup → ongoing review. I built the full stack, with campaigns created paused and a human approval step before launch. The interesting part: deciding where the agent has to stop.
Next.js FastAPI LangGraph Google Ads API
ZenAI — ask your database a question
A text-to-SQL and visualization assistant I led for Aqore through Mandala Labs. Retrieval checks, SQL validation and repair, streamed answers, and model routing. Because a query that runs successfully can still answer the wrong question.
Angular FastAPI PostgreSQL PydanticAI
FlashNotes — give your documents a chat box
An earlier project bringing document chat, summaries, quizzes, and OCR together with a local Qwen model. Public source and a demo recording.
Next.js FastAPI Qwen RAG
AlgoMaze — watching algorithms get lost
Generate a maze and watch A*, BFS, and DFS find their way through it. A small Python project for when I want to see what the algorithm is thinking.
Python Pygame
Python and TypeScript · React / Next.js · FastAPI · PostgreSQL · Docker
LangGraph · PydanticAI · local models · MCP · browser automation
I care about inspectable failures, useful tests, and knowing what an agent is allowed to do.
Tinkering with a Raspberry Pi desk pet. Apparently a pixel cat in my portfolio wasn't enough. 🐈
Open to full-stack AI and automation work. Tell me what you're building.