Local Browser AI Agents: Web-LLM in Delta Dynamics
Bringing client-side LLM-driven AI agents into the browser with WebGPU, MLC Web-LLM, and web workers for decentralized ecosystem simulation.
Numerical, discrete-event, and agent-based simulation models replicating real-world physical and operational phenomena.
Bringing client-side LLM-driven AI agents into the browser with WebGPU, MLC Web-LLM, and web workers for decentralized ecosystem simulation.
Real-time shallow water and groundwater simulation shaders on the GPU at 60 FPS using WebGL, Three.js, and FBO ping-pong texture computation.

Here is another interesting piece of work I was doing over the past three years with Dominik Ascher: Integrated transportation and power systems modeling to hel...

Besides our original topics such as distributed and smart fride control we have worked on another interesting issue: Smart traffic control. By smart traffic con...

To better understand the behavior space and its heuristic exploration using our tool Xtream we developed a proprietary 3D visualization. The visualization shows...

As part of our virtual smart grid testbed lab we have been working on an interactive visualization for our models including their structure and behavior. Here a...
Ecosystem Simulator
Low-poly ecosystem simulator featuring real-time dynamic terrain, GPU-accelerated water flow, resource management, and local LLM-driven AI behaviors.
Master-level course on Digital Factory and Computer Simulation (including practical exercises with the discrete event simulation software JaamSim).