Where AI meetsBayesian statistics
We build fast Bayesian methods for the age of AI that transcend disciplines and live beyond the words of a paper.
Ideas, tools, and people
Our work spans methodological research, open-source infrastructure, and collaborations across scientific domains.
Latest news
October 2026
- Konstantina Sokratous from the University of Missouri and Yufei Wu from KU Leuven visited BayesOps.
- Stefan T. Radev gave an invited talk on simulation-based inference at the Data-Driven Modeling, Simulation and Inference for Neurobiological Problems workshop hosted by NITMB.
September 2026
- Stefan T. Radev taught simulation-based inference and BayesFlow at Heidelberg University's AI for Science Summer School.
- Stefan T. Radev led the PyMC Labs webinar Hands-On Amortized Bayesian Inference: Train and Deploy Generative AI for Fast Bayesian Modeling.
- Stefan T. Radev received an NIH R01 award to advance fNIRS with simulation, generative AI, and Bayesian inference, in collaboration with the Intes Lab.
- Shishir Shrestha joined BayesOps as a PhD researcher. Welcome to the lab, Shishir!
August 2026
- Stefan T. Radev held four sessions on simulation-based inference as part of the Model-Based Neuroscience and Cognition Summer School at the University of Amsterdam.
- Jonas Arruda taught simulation-based inference at the Probabilistic AI School hosted by the Artificial Intelligence Association of Lithuania.
July 2026
- Stefan T. Radev co-organized the first summer school on Simulation-Based Inference for Cognitive Modeling at Ohio State University (OSU) sponsored by the William K. and Katherine W. Estes Fund.
- Stefan T. Radev organized a symposium on deep learning for computational cognitive modeling as part of MathPsych 2026.
- Jerry Huang, Niek Stevenson, Lukas Schumacher, and Mischa von Krause presented at MathPsych 2026.
May 2026
- Our short paper on inverting foundation brain models was accepted at the Structured Probabilistic Inference & Generative Modeling workshop at ICML 2026. Congratulations to Niels Bracher!
- Our papers on amortized Bayesian adaptive design and meta-amortized Bayesian inference were accepted at ICML 2026 and ProbML (oral)! Congratulations to Niels Bracher and Jerry Huang!
- Stefan T. Radev and Jonas Arruda presented at the Approximately Bayes Workshop organized by the International Centre for Mathematical Statistics.
What we build
Generative AI for Bayesian Statistics
BayesFlow
BayesFlow is a Python library for simulation-based inference with generative AI. It provides users and researchers with a user-friendly API for amortized Bayesian workflows, AI research, and multi-backend support via Keras3.
Model-based fNIRS analysis with AI
fNIRS SimBox
We are developing a full-head 3D fNIRS simulator that will be used for in silico neuroimaging and model-based analysis. We are also creating AI models for real-time model calibration and data interpretation, targeting application areas like surgical training.
Learn All Your Models Once
CogFormer
CogFormer is the first incarnation of the BayesGPT project, allowing cognitive modelers to perform near-instant Bayesian modeling on a combinatorial space of design configurations. The web app lets you interact with a pre-trained version of CogFormer v1.