MSc Software Engineering @ UE Potsdam · Relocating to Berlin · Open to Werkstudent roles.
I build things at the intersection of LLM agents and mobile. Most of my work involves wiring together LangGraph pipelines, FastAPI backends, and Flutter frontends into something that solves a real problem — not just a demo.
Right now I'm focused on:
- Multi-agent systems with LangGraph, RAG, and MCP
- Flutter apps that talk to AI backends
- Keeping infra free (Fly.io, Supabase, Groq, Vercel)
I'm not trying to do everything. I'm trying to be genuinely useful at the AI + mobile layer.
| AI & LLM | Frontend | Backend & DB | DevOps & Tools |
|---|---|---|---|
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Runs every day via GitHub Actions without me touching it. Searches Tavily for AI roles in Berlin, scores them with Llama 3 on Groq, then pushes results to Notion and sends a Gmail digest. Built it because manually scanning job boards was tedious and I wanted to see if I could automate the whole loop. Turns out you can.
Telegram bot backed by LangGraph + ChromaDB. Has persistent memory, real-time web search, and a human-in-the-loop approval step before it takes any action. The HITL part was the interesting design problem — figuring out when to stop and ask vs when to just proceed.
AI-powered German learning app. Built it partly because I'm moving to Germany and needed something better than Duolingo for reading actual documents. PDF scanning with Groq + Shadow Listening feature + offline-first via IndexedDB.
Zero-cost multi-agent productivity system. LangGraph orchestration, live web research, database task management. Runs entirely on free-tier infrastructure.
Merging MCP with native Flutter — most agent UIs are still web-only and I think there's something interesting in bringing tool-calling into mobile. Also finishing up a receipt scanner with Gemini Vision + Google Sheets that a few small businesses have asked about.
If you're building something at the AI + mobile intersection, I'm always up for a conversation.