Yuntian Deng
Assistant Professor at Waterloo Data & Artificial Intelligence Institute
Waterloo, Ontario, Canada
Summary
🤩
Rockstar🎓
Top SchoolYuntian Deng is an Assistant Professor and machine learning researcher with 11 years of experience bridging academic research and open-source engineering in NLP and multimodal deep learning. Educated at Tsinghua, CMU and Harvard (PhD), he has held research roles at AI2, Facebook AI Research, NVIDIA, and Bloomberg, and now splits time between the University of Waterloo and Harvard-affiliated work. His contributions to high-profile open-source projects like OpenNMT and the Harvard im2markup repo show hands-on expertise in model training, data preprocessing, and adapting sequence models to image and audio modalities. He combines rigorous research—evident from postdoctoral and visiting positions—with practical engineering improvements such as logging, memory optimization, and dataset pipelines. Colleagues know him for smoothing the path from prototype to reproducible code, and for quietly improving evaluation and preprocessing tools that make large-scale experiments possible.
11 years of coding experience
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Harvard UniversityBachelor of Science (BS), Automation, Bachelor of Science (BS), Automation at Tsinghua UniversityMaster's degree, Computer Science, Master's degree, Computer Science at Carnegie Mellon University
Github Skills (21)
pytorch10
machine-translation10
language-model10
preprocessing10
neural-machine-translation10
python10
evaluation10
image-processing10
preprocess10
torch-hub10
machine-learning10
lua10
deep-learning10
data-processing10
torch-net10
Programming languages (8)
Github contributions (5)
harvardnlp/im2markup
Sep 2016 - May 2022
Neural model for converting Image-to-Markup (by Yuntian Deng yuntiandeng.com)
Role in this project:
ML Engineer
ML EngineerContributions:17 commits, 2 PRs, 27 pushes in 5 years 8 months
Contributions summary:Yuntian appears to be making changes related to preprocessing scripts and model evaluation within an image-to-markup project. Their contributions include updates to `preprocess_latex.js` and `preprocess_formulas.py`, suggesting they're refining data preparation steps for LaTeX formulas. They've also modified `evaluate_image.py`, indicating involvement in the image evaluation process. Additionally, there are updates to the downsampling function.
deep-learningcomputer-graphicsconvertingimage-processingmarkup
OpenNMT/OpenNMT-py
Nov 2017 - Oct 2018
Open Source Neural Machine Translation and (Large) Language Models in PyTorch
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:96 commits, 26 PRs, 18 pushes in 11 months
Contributions summary:Yuntian contributed to the training process, modifying the `onmt/IO.py`, `onmt/ModelConstructor.py`, `train.py`, `onmt/Trainer.py`, `onmt/Translator.py`, and `preprocess.py` files. They appear to have worked on integrating image and audio data types, expanding the dataset processing capabilities and adapting the model for these data modalities. Their commits indicate changes related to model architecture, data loading, and the overall training workflow of the neural machine translation model.
pytorchdeep-learningmachine-translationtranslationneural-machine-translation
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.