whoami
Dimitrije Marković
cat about.md
Computational researcher and ML consultant. PhD in theoretical physics (Goethe Frankfurt, 2013), more than a decade as a postdoc at TU Dresden (2014–2025), then several years consulting on probabilistic and active-inference-based AI.
Industry work spans financial modelling, remote sensing, precision agriculture, and biomedical signals — data pipelines, probabilistic models, and proof-of-concept demos that have supported client fund-raising.
cat now.md
2026 · starting a company with co-founders; currently consulting for an early-stage startup. Open-source work on probabilistic programming in JAX (pybefit, bmr4pml, sppcax, bllarse).
Open to further engagements if the work is interesting.
ls --sort=year | head -5
-
2024 Bayesian sparsification for deep neural networks with Bayesian model reduction
-
2024 Gradient-free variational learning with conditional mixture networks
-
2024 Modelling agency perception in a multi-agent context in depression using active inference
-
2022 Revealing human sensitivity to a latent temporal structure of changes
-
2021 An empirical evaluation of active inference in multi-armed bandits
→ full list on Google Scholar
contact
github/dimarkov · scholar · slides · active inference institute (talk)