Right now I'm a Software Developer at snacc-it GmbH, contributing to Project ETHICA, a €6.2M EU-funded R&D initiative building a Digital Product Passport platform that uses LLMs and RAG to automate ESG compliance verification across global supply chains. Alongside that I work across ServiceNow: incident management, travel reimbursement, illness reporting, and enterprise web development. On the side I build AI projects like DeepScholar, a RAG research copilot that keeps its answers grounded in real, citable sources.
Before this I worked as a quantitative developer building trading bots, tutored 200+ students in computer science at FH Aachen, and co-founded a SMMA agency and a fitness clothing e-commerce brand, together generating €6,500+ in revenue. I write about what I learn on Medium, usually about building AI products people can actually trust.
Aachen, Germany
Remote-friendly
AI & ML focused
Trust by design
I'd rather ship AI that admits what it doesn't know. In DeepScholar, every answer cites a real passage or says it can't answer. No fabricated sources, no confident guesses.
Range across the stack
Next.js and FastAPI, Python ML pipelines, Java systems, ServiceNow scripting, even a trading bot. I follow the problem wherever it leads instead of staying in one lane.
The boring parts matter
Validation at every boundary, reproducible pipelines, and clean data the rest of a team can rely on. Most of what makes software trustworthy is unglamorous, and I lean into it.