Mojmir Mutny

Mojmir Mutny

Machine learning scienstist currently a postdoc at Broad Institute of MIT and Harvard. In the past I was at ETH Zurich working in Learning and Adaptive Systems group. My current research focuses on active learning and efficient fine-tuning, aiming to identify the most informative data to improve machine learning models. I work on improving the reliability of LLMs in generation and exploring new enzymes. I have a background in Bayesian optimization, kernel methods, sensor selection, control, and optimization. My algorithms have been applied to machine calibration, laboratory automation, spatial analysis, and enzyme and antibody design.

Open-source code packages

doexpy Image

Package 1: doexpy

A universal Active Learning (Design of Experiments) library using modern Machine Learning. Optimization of the most informative queries and policies, and much more!

stpy Image

Package 2: stpy

This code implements a general purpose stochastic process fiting, Gaussian process, Poisson Point processes, various emebeddings etc.

sensepy Image

Package 3: sensepy

This repository includes the code used in paper: Mojmir Mutny & Andreas Krause, “Sensing Cox Processes via Posterior Sampling and Positive Bases”, AISTATS 2022

QFF Image

Package 4: QFF

This repository includes the code used in paper: Mojmir Mutny & Andreas Krause, “Efficient High Dimensional Bayesian Optimization with Additivity and Quadrature Fourier Features”, NIPS 2018 It provides an efficient finite basis approximation for RBF and Matern kernels in low dimensions.