I start with the problem.
The tool follows.
Most AI engineers reach for an LLM first — I don't. Sometimes the right answer is an agent, sometimes a vector database, sometimes a classical model that runs in milliseconds and costs almost nothing. I'm the founding AI engineer at Tekkr, ex-Siemens (Energy, eMobility, Mobility). M.Sc. in Computer Engineering — ML & AI — from TU Berlin.
Most AI engineers reach for an LLM first. I don't.
I start with the problem and work backwards to the simplest system that solves it. Sometimes that's an agent. Sometimes a vector database. Sometimes a classical model that runs in milliseconds and costs almost nothing. The tool follows the problem.
- 01 · Problem
Sit with the problem first. The thing the team actually loses sleep over.
5 years of customer-facing work at Siemens — long before any code.
- 02 · Data
Most problems are data problems before they're model problems. Pipelines, observability, feedback loops.
Telemetry pipelines for thousands of EV chargers at Siemens AG.
- 03 · System
Pick the simplest thing that solves it — agent, vector DB, classical model, sometimes just SQL.
GraphRAG when structure matters. Logistic regression when latency matters.
- 04 · Ship
Ship it where real teams will use it. Reliability and cost matter as much as accuracy.
Production RAG platforms and internal AI tooling at Tekkr.
Things I've built end-to-end.
Research, full-stack products, systems work. Every flagship project ships with a live demo or a recorded walkthrough.
Teaching LLMs the shape of a codebase.
A two-stage retrieval system combining a Knowledge Graph of source code with embeddings. Built at Siemens Mobility for safety-critical codebases — reduced developer onboarding time by ~30%.
- 01~30% onboarding time reduction (Siemens Mobility user studies)
- 02Final grade 1.3 (sehr gut)
- 03Two-stage RAG: vector search + graph traversal
- 04Multi-language parsing (Python, C, C++) via Tree-sitter
- 05Resumable enrichment that survives API rate limits
- 06Scales to 30k+ file repositories
A short version.
The long version lives at /about.
I'm an AI engineer based in Berlin, founding engineer at Tekkr. I build production AI systems — RAG platforms, agents, internal tooling — for teams that need them to actually work.
Before Tekkr: five years in technical sales at Siemens Energy (which taught me to sit with customers before writing a line of code), then data pipelines for thousands of EV chargers at Siemens AG (eMobility), and my M.Sc. thesis at Siemens Mobility building a GraphRAG assistant for safety-critical codebases.
M.Sc. Computer Engineering with emphasis in ML & AI from TU Berlin (thesis: 1.3 — sehr gut). C/C++ at 42 Berlin to keep the systems intuition sharp.
- RoleFounding AI Engineer @ Tekkr
- BackgroundEx-Siemens (Energy / eMob / Mobility)
- M.Sc.Computer Engineering · ML & AI · TU Berlin
- B.Sc.Electrical Engineering · TU Berlin
- Thesis1.3 (sehr gut)
- Systems42 Berlin · C/C++
- FocusProduction ML · RAG · Agents
- LocationBerlin
- StatusOpen to interesting collaborations
Got something to build?
I care about systems that are simple, reliable, and actually used. Happy to talk about RAG, agents, classical ML, and product decisions that come before any of it.