📍 Geneva, Switzerland | 🔬 Systems researcher and entrepreneur | 🎓 EPFL → MIT → MPI-SWS
I lead the Data Systems Group at MPI-SWS. We work across systems, databases, and machine learning, and most of what we build now is infrastructure for agentic AI: systems that coordinate many models over data far larger than any one of them.
- 🛡️ scode – Sandbox wrapper that keeps AI agents away from your credentials
- 🎛️ codemux – One CLI for thirteen AI coding agents, every run sandboxed by scode
- 📊 usagemux – Subscription quota and credits, normalized across providers
See also: mpi-dsg for more research prototypes.
- 🌊 X-Stream – Edge-centric graph processing using streaming partitions
- 🌀 Chaos – Scale-out graph processing from secondary storage
- ⛈️ Hailstorm – Disaggregated compute and storage for distributed LSM-based databases
- 🌪️ Hurricane – Taming skew in large-scale analytics
- 🔮 Tesseract – Tesseract implementation in Python
- 🗃️ Loki – Generate databases from high-level workload statistics
- Agentic AI Infrastructure – Building systems that let AI agents operate at scale
- Cache Saver – LLM inference optimization framework
- Code Intelligence – SQL-of-Thought and LoRACode, teaching models to reason about code
- LLM-Enhanced Databases – Rethinking query processing with learned components
- Teaching – Operating Systems, Distributed Systems, and Systems for LLMs at Saarland
"Understand the system before you optimize it."
I build data systems that scale and teach machines to reason about code. The best research ships as real software.
- I have moved between research and startups my whole career
- PhD at EPFL, postdoc at MIT CSAIL, now group leader at MPI-SWS
- Swiss Armed Forces veteran (IT Pioneer / Chief of Staff)
- I work on systems that end up running in production, not only in papers