Marius Pachitariu
Lab Head at Howard Hughes Medical Institute (HHMI)
Ashburn, Virginia, United States
Summary
🤩
Rockstar🎓
Top SchoolMarius Pachitariu is a lab head at HHMI Janelia with a decade of experience at the intersection of neuroscience and machine learning, leading a research group that develops tools for neural data analysis. He holds a PhD in Computational Neuroscience and Machine Learning from UCL and a BA in Mathematics from Princeton, and previously completed a postdoc in the Harris/Carandini lab. Marius is a prolific open-source contributor—known for work on influential projects like Kilosort (CUDA-optimized spike sorting) and Cellpose—bringing deep algorithmic insight and GPU-level optimizations to production-grade neuroscience software. His contributions span both foundational research and practical tooling, from spike-detection kernels to deep-learning notebooks for neuroimaging, reflecting a rare blend of mathematical rigor and systems-level engineering. Based in Ashburn, VA, he combines hands-on coding with lab leadership, prioritizing scalable, reproducible workflows that accelerate discovery.
10 years of coding experience
3 years of employment as a software developer
University College LondonBachelor of Arts, Mathematics, Bachelor of Arts, Mathematics at Princeton University
Github Skills (17)
algorithm10
algorithms10
convolutional-neural-networks10
c-language10
electrophysiology10
data-structure10
neuroscience10
deep-learning10
cuda10
data-structures10
computer-vision10
jupyter-notebook10
cprogramming-language10
image-classification9
pytorch9
Programming languages (4)
Github contributions (5)
MouseLand/Kilosort
May 2018 - Jan 2023
Fast spike sorting with drift correction
Role in this project:
Backend Developer
Backend DeveloperContributions:14 releases, 1 review, 346 commits in 4 years 9 months
Contributions summary:Marius made numerous changes to CUDA files related to spike sorting, specifically focusing on algorithmic improvements and optimizations within the Kilosort project. Their work primarily involved refining core functions such as `computeCost`, `bestFilter`, `average_snips`, and other critical components related to spike detection, clustering, and feature extraction. These changes suggest a deep understanding of the underlying mathematical and computational aspects of spike sorting, and their contributions likely enhanced the performance and accuracy of the Kilosort algorithm.
neurosciencerustspike-sortingsortingchannels
NeuromatchAcademy/course-content-dl
Jul 2021 - Jul 2022
NMA deep learning course
Role in this project:Data Scientist
Data ScientistContributions:15 reviews, 36 commits, 77 PRs in 1 year
Contributions summary:Marius's commits primarily involve modifications to Jupyter notebooks related to deep learning. They explore and implement deep learning techniques such as animal pose estimation, image classification, and sequence-to-sequence models for neuroscience. The user's work also involves fine-tuning CNNs on fMRI data and analyzing spectrograms. The user's contributions contribute to learning and applying deep learning methods within the scope of the repository's goals.
reinforcement-learning-algorithmstransformersdeep-learningrecurrent-neural-networkscontinual-learning
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.