Charles Martin
Founder at Calculation Consulting
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolCharles Martin is a founder and AI consultant with 13+ years building and shipping machine learning and deep learning systems across search relevance, NLP, recommendation, and quantitative finance for firms like GLG, BlackRock, eBay, GoDaddy and Demand Media. He blends theoretical rigor—a PhD in theoretical chemical physics and NSF fellowship—with hands-on full-stack engineering (Python, Java, Ruby, C/Fortran) and production ML at scale. Charles leads the open-source WeightWatcher project for predicting DNN accuracy, contributing novel metrics like matrix entropy and ONNX support, and collaborates with UC Berkeley on AI foundations. He has a track record of converting research into commercial impact (e.g., models behind Demand Media’s content valuation and Aardvark work that preceded a Google acquisition). Comfortable from Cray supercomputers to AWS/DevOps, he also advises the Page family office on advanced physics and climate-focused technologies.
13 years of coding experience
2 years of employment as a software developer
NSF Postdoctoral Fellow Neural Networks and Theoretical BioPhysics, NSF Postdoctoral Fellow Neural Networks and Theoretical BioPhysics at University of Illinois Urbana-ChampaignPhD & NSF Fellow Theoretical Chemical Physics, PhD & NSF Fellow Theoretical Chemical Physics at University of ChicagoB.S Math Chemistry, B.S Math Chemistry at The Ohio State University
Github Skills (8)
machine-learning10
deep-learning10
python10
svd10
modeling10
onnx9
tensorflow9
pytorch9
Programming languages (5)
Github contributions (5)
CalculatedContent/WeightWatcher
Nov 2018 - Jan 2023
The WeightWatcher tool for predicting the accuracy of Deep Neural Networks
Role in this project:
ML Engineer
ML EngineerContributions:10 releases, 790 commits, 39 PRs in 4 years 2 months
Contributions summary:Charles implemented significant features for analyzing Deep Neural Networks by creating tools for predicting the accuracy of models. Their work involved analyzing the weight matrices of Deep Neural Networks (DNNs). The contributions also included implementing new metrics such as the matrix entropy and modifications to existing tools to improve their performance and overall functionality. The user also added functionality for support of different ONNX models.
deep-learningneural-networksmachine-learningneural-networkdeep-neural-networks
CalculatedContent/big_deep_simple_mlp
Apr 2016 - Jul 2016
Contributions:83 commits, 1 PR, 81 pushes in 2 months
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.