Haokun Liu
Research Assistant at Vector Institute
Canada
Summary
🤩
Rockstar🎓
Top SchoolHaokun Liu is a research-focused machine learning engineer with seven years of experience building and evaluating NLP models across academia and industry, currently a Research Assistant at the Vector Institute in Canada. His background includes research scientist roles at IBM and Adaptive ML, student research positions at Google, and hands-on contributions at NYU and AI2, reflecting a strong bridge between cutting‑edge research and practical engineering. He has a solid academic foundation including advanced work toward a PhD in Computer Science and an MS in NLP, and a track record improving widely used tooling—contributing bug fixes, refactors, tests, and task additions to the popular jiant NLP toolkit. Haokun’s strengths lie in model evaluation, dataset/task integration, and production-ready code quality, and he often focuses on non-obvious areas like annotation issues and vocabulary interoperability that materially improve downstream benchmarking.
7 years of coding experience
8 years of employment as a software developer
Master of Computer Science, Natural Language Processing, 3.83, Master of Computer Science, Natural Language Processing, 3.83 at New York UniversityBachelor of Science - BS, Computer Intelligence Science, 3.85, Bachelor of Science - BS, Computer Intelligence Science, 3.85 at Peking UniversityDoctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of TorontoDoctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of North Carolina at Chapel Hill
Github Skills (9)
transformers10
pytorch10
nlp10
bert10
python9
multi-task-learning9
pandas8
unit-testing7
machine-learning7
Programming languages (3)
Github contributions (5)
nyu-mll/jiant
Mar 2019 - Jan 2021
jiant is an nlp toolkit
Role in this project:
ML Engineer
ML EngineerContributions:3 reviews, 86 commits, 28 PRs in 1 year 10 months
Contributions summary:Haokun primarily contributed to the development and improvement of the `jiant` NLP toolkit. Their work included bug fixes related to BERT, refactoring code to use pandas and numpy for data loading, and adding unit tests. Furthermore, they integrated AllenNLP vocabulary and addressed issues related to task configuration and data annotation, especially for the CoLA analysis tasks and the addition of HellaSwag and Winogrande tasks, reflecting their focus on model evaluation and performance improvements.
nlptransformersmultitask-learningsentence-representationbert
nyu-mll/wsc-formalizations
Mar 2020 - Oct 2020
Contributions:100 commits, 2 PRs, 12 pushes in 7 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.