(Cologne AI)

Human-machine interaction through natural language.
Christoph Schütte + Mirko Klukas


We are two tech founders in the early stages of the founding process combining research and industry experience.

Our roots lie in Cologne, Germany, where we went to university, drank Kölsch, and reminisced about life and mathematics. We always shared a deep fascination for science and technology which led us on individual journeys through MIT, Google, and different startups working on a broad set of intricate problems. In the aftermath of the pandemic we returned to our roots and teamed up to build something new.

Current Focus, and
Latest Posts

Our current work focuses on natural language processing (NLP), transformer-based large language models (LLMs), and information retrieval. However, we are excited about deep learning, and symbolic approaches alike.

Follow our progress:

Speaker segmentation in Gen — Probabilistic modeling of speaker similarity matrices 15 Oct 2022
Denoising Diffusion Model — A minimal example generating MNIST images 14 Oct 2022
Deriving the simplified training loss for diffusion models — Bypassing ELBO and KL-divergences 01 Oct 2022
The Metropolis-Hastings Algorithm — MCMC Snippets (1) 15 Sep 2022
The Chinese Restaurant Process — Derivation from finite Dirichlet mixtures 01 Sep 2022