Selected work / product engineering
My work spans product decisions, interfaces, integrations, and the details that make a workflow reliable.
Start with the product I built myselfIndependent product / 2026–present
A commerce ERP for everything that happens after a marketplace order arrives.
A seller works across roughly three to four marketplace stores and two to four shipping integrations. Orders, stock, packing, shipping, returns, and accounting need to remain connected.
I designed and implemented the product end to end, from role-specific screens and order handling to marketplace and carrier integrations. It is now used for the seller’s operations.
The public sandbox uses fictional records and Shopify and DHL test environments. Choose a role on the product page, then follow a real workflow.
The reason and next decision are visible next to the affected order.
Selected work in teams
These projects were team efforts. The notes below focus on my contribution and what changed, without reproducing employer screens or code.
Enterprise architecture reporting
Made complex report data easier to explore: relationship views, counts across nested clusters, GraphQL data paths, and controls that let an AI assistant change real report settings.
Autonomous-driving data tools
Built Python services and JavaScript interfaces for data retrieval and map workflows. An R-tree spatial indexing approach tripled retrieval speed and reduced manual annotation by 75% in the workflow I worked on.
Research operations software
Built an ASP.NET and Vue application for concurrent conversion jobs and metadata retrieval. The team measured an 800% improvement in conversion time over the earlier solution.
Research / University of Bonn
For my master’s thesis, I compared entity-guided retrieval with conventional approaches to financial question answering. The useful lesson carried into product work: establish a baseline, inspect failures, and make claims the evidence supports.
Want the complete chronology and skills?
Read the CV