About
I'm a Machine Learning Researcher at ElevenLabs working on geneartive audio models. Previously I studied Machine Intelligence (MSc) and Computer Science (BSc) at ETH Zürich.
I created the archinet GitHub group with the goal of replicating papers results and publishing useful ML-related libraries. The philosophy of Archinet is to keep everything as simple as possible, modular, and easy to understand. Some popular projects include surgeon-pytorch, a library to inspect and extract parts of PyTorch models, and audio-diffusion-pytorch, a set of diffusion models for audio generation I worked on during my Master's thesis, which resulted in Moûsai: Text-to-Music Generation with Long-Context Latent Diffusion.
For fun, I like to do web-related stuff (development/design), like this blog. The current stack is React.js with Next.js for server-side rendering, Stitches for component styling, and MDX.js to blog with markdown syntax.