Jesse Chan
United States
Summary
🤩
Rockstar🎓
Top SchoolJesse Chan is a pragmatic software engineer with seven years of experience building ML-backed backend systems and production-ready models, currently at Hudson River Trading. He has shipped features at Google on the Keras/TensorFlow stack—working on LLM components like GPT, BERT and RoBERTa—and contributed tests and robustness fixes to notable open-source projects such as Keras and MLflow. His background spans full-stack backend work at SeekOut (image classification, big-data pipelines, recommendation systems), mobile and C++ vision optimizations at Clobotics, and hands-on MLOps improvements to image logging in MLflow. A former CMU teaching assistant who co-founded a Python bootcamp, he combines clear technical communication and mentorship with a taste for practical, well-crafted software. Notably, his contributions emphasize model reliability (checkpointing and decoder tests) and edge-case handling in telemetry—skills that help bridge research-quality ML with production constraints.
7 years of coding experience
2 years of employment as a software developer
International Baccalaureate, High School Diploma, International Baccalaureate, High School Diploma at Shanghai American SchoolBachelor of Science - BS, Computer Science, 3.96, Bachelor of Science - BS, Computer Science, 3.96 at Carnegie Mellon University
Chinese, English, Chinese
Github Skills (17)
transformers10
python10
image-processing10
machine-learning10
mlflow10
numpy10
keras10
mlops10
tensorflow10
transformer10
nlp10
testing9
pytest9
ai8
ml7
Programming languages (4)
Github contributions (5)
keras-team/keras-hub
Jun 2022 - Sep 2022
Pretrained model hub for Keras 3.
Role in this project:
ML Engineer
ML EngineerContributions:51 reviews, 13 commits, 22 PRs in 2 months
Contributions summary:Jesse's contributions primarily focus on modifying and testing the `TransformerDecoder` layer within the KerasNLP framework. Their work includes adding optional parameters, refining initialization, and addressing issues identified in pull requests. The changes involve implementing tests to ensure the `TransformerDecoder` functions correctly with and without cross-attention, demonstrating a focus on functionality and robustness of the model. Furthermore, the user implemented tests and checkpointing for the model, showcasing a deep understanding of the RoBERTa model.
nlppythondeep-learninglanguage-processingmachine-learning
mlflow/mlflow
Sep 2023 - Jul 2024
Open source platform for the machine learning lifecycle
Role in this project:Back-end Developer & MLOps Engineer
Back-end Developer & MLOps EngineerContributions:44 reviews, 42 PRs, 12 pushes in 9 months
Contributions summary:Jesse made contributions focused on improving the MLflow's image logging functionality. The user updated the `mlflow.client.py` to support auto-scaling float values within the images. Additionally, the user updated `mlflow/tracking/client.py`, `mlflow/tracking/multimedia.py`, and related test files, demonstrating skills in the core functionality of MLflow's tracking capabilities. The changes show a focus on enhancing the logging of model evaluation results by adding support for logging image and handling edge cases.
pythonlifecyclemlmachine-learningincremental-learning
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.