Sam Bowman
Member Of Technical Staff at NYU AI School
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolSam Bowman is a researcher-engineer who studies neural models for natural language understanding and applies them to questions in syntax and semantics, blending rigorous academic training (PhD, Stanford; BA/MA, Chicago) with 11 years of engineering experience. Currently a Member of Technical Staff at Anthropic leading AI safety and evaluation research teams, he remains an associate professor on leave at NYU and actively contributes to NLP tooling—most notably backend model loading, checkpointing, and evaluation fixes in the widely used jiant toolkit. Based in San Francisco, he focuses on high-quality training and evaluation data and pragmatic model engineering to ensure robust experimental results. Known for clear boundaries about industry moves, he prefers scholarly and public-facing engagement (personal site and Twitter) over recruitment, and combines deep linguistics insight with production-grade ML systems design.
11 years of coding experience
3 years of employment as a software developer
BA/MA Linguistics, BA/MA Linguistics at University of ChicagoPhD Dept. of Lingistics (NLP Group), PhD Dept. of Lingistics (NLP Group) at Stanford University
Github Skills (10)
pytorch10
nlp10
checkpoint10
model-checking10
evaluation10
python10
checkpointing10
bert9
transformers9
multi-task-learning8
Programming languages (6)
Github contributions (5)
nyu-mll/jiant
Jun 2018 - Oct 2020
jiant is an nlp toolkit
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:6 releases, 9 reviews, 140 commits in 2 years 3 months
Contributions summary:Sam primarily contributed to the model's checkpointing and evaluation logic within the "jiant" NLP toolkit. They focused on ensuring that the correct model checkpoints from pretraining and target training runs were loaded and utilized during evaluation. Furthermore, they implemented error handling to prevent the evaluation of untrained models and made fixes related to ELMo model saving. The user demonstrated expertise in model loading, and also contributed to the results script.
nlptransformersmultitask-learningsentence-representationbert
RepEval2017/RepEval2017.github.io
Feb 2017 - Apr 2018
Contributions:25 commits, 1 PR, 51 pushes in 1 year 2 months
javascriptreactjekyllworkshop
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.