John Giorgi
Senior ML Research Scientist at Abridge
Canada
Summary
👤
Senior🎓
Top SchoolJohn Giorgi is a Senior ML Research Scientist with 11 years of experience bridging academic research and production ML, currently leading post-training efforts at Abridge after earning a PhD in Computer Science from the University of Toronto. He has a strong track record in NLP and model engineering, contributing substantive fixes and training optimizations to the high-profile AllenNLP ecosystem, including AMP support and sequence-to-sequence improvements like label smoothing and scheduled sampling. His background includes multiple research internships at the Allen Institute for AI and applied projects in health tech and web mining, reflecting a focus on real-world language and healthcare applications. Comfortable moving models from research into robust pipelines, he blends deep theoretical training with hands-on engineering to improve training stability and data processing. A former biochemistry undergraduate with lab research experience, he brings an unusual cross-disciplinary perspective to ML problems involving biomedical and clinical text.
11 years of coding experience
2 years of employment as a software developer
Doctor of Philosophy, Computer Science, Doctor of Philosophy, Computer Science at Department of Computer Science, University of TorontoBachelor of Science (B.Sc.), Major in Biochemistry / Minor in computer science, Bachelor of Science (B.Sc.), Major in Biochemistry / Minor in computer science at University of Ottawa
Github Skills (22)
beam-search10
pytorch10
python10
deep-learning10
computer-engineering10
natural-language-processing10
sequence-to-sequence10
nlp10
data-science9
system-configuration9
data-processing9
custom-configuration9
configurations9
yml-configuration9
pytest8
Programming languages (14)
Github contributions (5)
allenai/allennlp
Nov 2019 - Jan 2022
An open-source NLP research library, built on PyTorch.
Role in this project:
ML Engineer & Software Engineer
ML Engineer & Software EngineerContributions:12 reviews, 20 commits, 22 PRs in 2 years 2 months
Contributions summary:John primarily contributed to the AllenNLP library by fixing bugs, improving code quality, and enabling new functionalities related to training and model loading. They addressed issues with f-string formatting and type hints, improving code readability and maintainability. Their contributions also included enabling Automatic Mixed Precision (AMP) training with Apex and implementing features for persisting and loading AMP state, indicating involvement in optimizing model training. Furthermore, they worked on addressing pickle bugs in dataset readers, showcasing their knowledge of data processing.
natural-language-processingpytorchnlpdeep-learningdata-science
allenai/allennlp-models
Nov 2020 - Dec 2021
Officially supported AllenNLP models
Role in this project:ML Engineer
ML EngineerContributions:14 reviews, 7 commits, 8 PRs in 1 year 1 month
Contributions summary:John primarily contributed to improving and extending the AllenNLP models, with a focus on the generation tasks. They addressed issues related to tokenization, start/end symbols, and beam search within sequence-to-sequence models. Furthermore, the user introduced and integrated label smoothing and scheduled sampling in CopyNet, enhancing its training capabilities. Additionally, the user worked on updates and configurations for the model, ensuring backward compatibility.
allennlpnlppytorch
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.