Joseph Cohen
Applied Scientist at Academic Torrents
Greater Seattle Area United States
Summary
🤩
Rockstar🎓
Top SchoolJoseph Cohen is an applied scientist and founder with 14 years of experience building ML systems at the intersection of medical imaging, reproducible data sharing, and production-scale cloud services. He has led research and product work at AWS, Stanford AIMI, Mila, and Butterfly Network, producing patented solutions for imaging workflows and migrating LLM/text services into PyTorch for HealthLake. As founder/director of Academic Torrents and contributor to high-impact medical repos like torchxrayvision and the COVID chest X-ray dataset, he blends open-science infrastructure with hands-on dataset curation and model engineering. His background spans startups, large tech, and academia, securing multi‑million grant funding and shipping accessibility and healthcare products such as BlindTool and portable ultrasound features. Uncommonly, he pairs deep research (PhD) with operational instincts—acquiring donated hosting resources to sustain a national-scale academic data distribution nonprofit.
13 years of coding experience
14 years of employment as a software developer
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at UMass BostonMassachusetts Bay Community College
Github Skills (20)
pytorch10
python10
data-science10
image-classification10
pandas10
machine-learning10
datasets10
vi10
deep-learning10
medical-imaging10
computer-vision10
co10
covid-testing10
data-analysis10
faster-rcnn9
Programming languages (14)
Github contributions (5)
mlmed/torchxrayvision
Mar 2020 - Jan 2023
TorchXRayVision: A library of chest X-ray datasets and models. Classifiers, segmentation, and autoencoders.
Role in this project:
ML Engineer & Data Scientist
ML Engineer & Data ScientistContributions:17 releases, 17 reviews, 325 commits in 2 years 11 months
Contributions summary:Joseph's contributions primarily focused on developing and maintaining the `torchxrayvision` library, specializing in chest X-ray datasets and models. The commits show the user working on adding, refactoring, and fixing dataloaders for various datasets, including NIH, PC, CheXpert, and COVID-19 data. They also worked on adding the ability to apply masks, implementing a PSPNet pre-trained model, and refactoring the training script, all geared towards enabling the use of the library for image classification and segmentation.
medical-applicationimage-classificationmedicalchest-radiographschest-xray-images
ieee8023/covid-chestxray-dataset
Feb 2020 - Jan 2021
We are building an open database of COVID-19 cases with chest X-ray or CT images.
Role in this project:Data Scientist
Data ScientistContributions:5 releases, 348 commits, 71 PRs in 11 months
Contributions summary:Joseph primarily focused on analyzing and modifying the dataset related to COVID-19 chest X-ray images. Their work included adding a test script for the dataloader, updating the test script to output statistics, and removing unusable X-ray images. Furthermore, the user added a script to generate an HTML database for visualizing the data. These efforts suggest a focus on data quality and exploration.
casesraycomputed-tomographydeep-learningdataset
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.