Mirko

Mirko Klukas, Ph.D.

based in Berlin, Germany Cambridge San Francisco Vienna Cologne

It's all just
Maps, Functions, and Arrows.

I always wanted to study art but became a mathematician. Contrary to common belief, it doesn't mean I simply like numbers. The beauty lies in-between, within their structure. Mathematics is about finding the right perspective and language to express a problem until the solution follows easily. It is a powerful way of thinking.

I am a research scientist and engineer working remotely for the Probabilistic Computing Group at MIT. I am specifically involved in the MIT Quest for Intelligence ChiSight perception moonshot. The topics I focus on within this project include Spatial AI, SLAM (Simultaneous Localization and Mapping), inverse graphics, and 3D scene understanding.

Prior to and during the pandemic, I had the opportunity to work in the United States in the fields of artificial intelligence and computational neuroscience. I was a researcher in Ila Fiete`s group at MIT's Department of Brain and Cognitive Sciences, and a research scholar at Numenta, a private research lab located in the San Francisco Bay Area. Before these experiences, I completed postdoctoral positions at the Institute of Science and Technology Austria and the University of Cologne, and even dabbled in the world of management consulting for a brief period.

The core of my research revolves around the integration of mathematics, neuroscience, and computer science. In recent years, my primary focus has been on investigating spatial representations within the brain and their significance in broader cognitive computations and the development of intelligence. My academic foundation however lies in pure mathematics, specifically in geometric and differential topology — with a particular emphasis on contact and symplectic topology. Believe me, it sounds more daunting than it truly is :) In fact, I miss it a lot.

Latest Posts

Speaker segmentation in Gen — Probabilistic modeling of speaker similarity matrices 15 Oct 2022
Diffusion models 101 — Approximate reverse kernel bypassing ELBO and KL-divergences 01 Oct 2022
The Chinese Restaurant Process — Derivation from finite Dirichlet mixtures 01 Sep 2022
The Metropolis-Hastings Algorithm — MCMC Snippets (2) 19 Jul 2022
Uniqueness of the stationary distribution — MCMC Snippets (1) 06 Feb 2022
— more posts —