Ross Girshick
Member Of Technical Staff at Anthropic
Seattle, Washington, United States
Summary
🤩
RockstarRoss Girshick is a seasoned machine learning researcher and engineer with 14 years of experience specializing in computer vision and object detection. He has held research and engineering roles at Meta FAIR, Microsoft Research, the Allen Institute, and now Anthropic, and co-founded a startup focused on applied ML. Ross is a principal contributor to landmark open-source projects such as R-CNN, Fast R-CNN, Faster R-CNN, Detectron and Detectron2, where his work on training pipelines, bounding-box regression, memory efficiency, and dataset support materially improved robustness and usability. His contributions span both research-grade algorithmic advances and pragmatic engineering—implementing efficient feature caching, multi-GPU testing support, and configuration tooling (yacs). Based in Seattle, he combines deep academic pedigree with production-facing software craftsmanship, and is known for quietly improving the plumbing that lets cutting-edge vision models scale.
14 years of coding experience
11 years of employment as a software developer
Github Skills (28)
pytorch10
caffe10
c-language10
python10
testing10
configuration-management10
machine-learning10
matlab10
detectron10
numpy10
deep-learning10
trainings10
retinanet10
object-detection10
computer-vision10
Programming languages (6)
Github contributions (5)
rbgirshick/voc-dpm
Jan 2012 - Apr 2017
Object detection system using deformable part models (DPMs) and latent SVM (voc-release5). You may want to use the latest tarball on my website. The github code may include code changes that have not been tested as thoroughly and will not necessarily reproduce the results on the website.
Role in this project:
Back-end Developer & Data Scientist
Back-end Developer & Data ScientistContributions:180 commits, 2 pushes, 3 comments in 5 years 3 months
Contributions summary:Ross implemented and refactored code related to feature vector caching, memory management and model training for a computer vision object detection system. The user's commits show code changes to support the training of a deformable parts model, along with modifications to improve memory efficiency, particularly for large datasets. The work also focused on the integration of a specialized bounding box prediction code that uses an external compiled library.
svmtarballobject-detectioncomputer-visionvoc
rbgirshick/rcnn
Mar 2014 - Apr 2017
R-CNN: Regions with Convolutional Neural Network Features
Role in this project:ML Engineer
ML EngineerContributions:80 commits, 4 pushes, 1 branch in 3 years 1 month
Contributions summary:Ross primarily contributed to the R-CNN project by addressing issues related to file paths for Caffe nets and updating the README file. They also added utility scripts for comparing feature cache files. Furthermore, the user made changes to incorporate copyright notices in various code files. A significant contribution involved the addition of a demo with an option to run it in CPU mode.
deep-learningregionsr-cnnconvolutional-neural-networkconvolutional
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.