Jacob Schreiber
Assistant Professor at Broad Institute of MIT and Harvard
Worcester, Massachusetts, United States
Summary
🤩
Rockstar🎓
Top SchoolJacob Schreiber is an Assistant Professor in Genomics and Computational Biology at UMass Chan with 12 years of experience applying large-scale machine learning to genomics. His work spans high-resolution 3D genome prediction from low-cost assays, deep tensor factorization for epigenomic imputation, and methods that combine expert knowledge with data to make Bayesian network structure learning tractable at scale. A former Stanford postdoc and IMP guest scientist, he brings strong systems and mathematical grounding from a PhD at the University of Washington. Jacob is also a seasoned open-source maintainer—he was a core developer on scikit-learn (notably improving numerical stability and ensemble APIs) and the author of pomegranate, a performant probabilistic modeling library. He blends rigorous research with practical software engineering, frequently shipping production-grade algorithms and tooling. Based in Worcester, he continues to bridge academic innovation and community-facing software in computational biology.
12 years of coding experience
9 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Computer Science, Doctor of Philosophy (Ph.D.) Computer Science at University of WashingtonUniversity of California Santa Cruz
Github Skills (17)
algorithm10
probabilistic-programming10
python10
data-science10
scikit10
data-modeling10
machine-learning10
machine-learning-algorithms10
probabilistic-reasoning10
numpy10
gradient-boosting10
statistical-models10
cython10
probabilistic-models10
bayesian-network10
Programming languages (12)
Github contributions (5)
jmschrei/pomegranate
Dec 2014 - Nov 2022
Fast, flexible and easy to use probabilistic modelling in Python.
Role in this project:
ML Engineer
ML EngineerContributions:5 releases, 4 reviews, 661 commits in 8 years
Contributions summary:Jacob's commits focused on implementing the sum-product algorithm in the `pomegranate/distributions.pyx` file, which involves integrating the algorithm with existing code. Their contributions involved modifications to conditional and discrete distributions within the library, which were focused on a Bayesian Network, with a likely focus on the math behind the algorithm. These changes appear to be part of a project that aims to improve and extend the capabilities of pomegranate's Bayesian network models.
pythonmachine-learningprobabilistic-graphical-modelspytorch
scikit-learn/scikit-learn
Jun 2015 - May 2017
scikit-learn: machine learning in Python
Role in this project:Data Scientist
Data ScientistContributions:10 commits, 52 PRs, 25 pushes in 1 year 11 months
Contributions summary:Jacob primarily contributed to the `scikit-learn` repository, focusing on the numerical stability and functionality of the machine learning algorithms. Their work includes adding an epsilon value to weight calculations in the NIPALS inner loop to improve numeric stability on Windows platforms. Furthermore, they modified the addition of the epsilon term to handle cases where y-weights are near zero, ensuring more robust model behavior. Additionally, the user refactored the code, splitting a Cython file into several parts and also added the apply method to Gradient Boosting in both Classifier and Regressor modules, enhancing the usability and functionality of the library.
machine-learningpythonscikit-learnstatisticsdata-science
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.