Josh Chang
Machine Learning Engineer at Nextdoor
Seattle, Washington, United States
Summary
👤
Senior🎓
Top SchoolJosh Chang is a machine learning engineer with eight years of experience, currently pursuing an M.S. in Computer Science at Columbia and working on applied ML at Nextdoor from Seattle. He combines solid research foundations from National Taiwan University and MIT-adjacent lab experience in NLP and GNNs with hands-on production work—contributing GCN implementations to TensorFlow's neural-structured-learning repository. His internships at Apple, Yahoo, Cathay Life, and IBM span recommendation systems, GPT-2 slogan generation, fraud detection with Neo4j+GNNs, and aggressive model compression via knowledge distillation. Josh’s strength is bridging research and engineering: he ships end-to-end models that integrate graph signals and large-scale data pipelines while squeezing practical gains (e.g., measurable accuracy and size improvements). He maintains a public portfolio of projects and publications that surfaces both his experimental work and production-ready engineering.
8 years of coding experience
Bachelor of Science, Computer Science & Information Engineering, Bachelor of Science, Computer Science & Information Engineering at National Taiwan UniversityMaster of Science, Computer Science, Master of Science, Computer Science at Columbia University in the City of New York
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (9)
neural-network10
keras10
tensorflow10
graph-neural-network10
python10
machine-learning9
adversarial-learning6
regularization5
graph4
Programming languages (1)
Github contributions (5)
tensorflow/neural-structured-learning
Jul 2020 - Dec 2020
Training neural models with structured signals.
Role in this project:
ML Engineer
ML EngineerContributions:19 reviews, 16 commits, 5 PRs in 5 months
Contributions summary:Josh's commits primarily involve implementing and modifying code related to graph neural networks within the `tensorflow/neural-structured-learning` repository. Their work includes defining and training a GCN (Graph Convolutional Network) model, including its layers, and integrating it with the dataset loading and training routines. They are also making changes in other files, showing that they understand the entire project workflow. The commits also involve code to test and predict with trained models.
adversarial-learningautoencodergraph-learningstructureddeep-learning
joshchang1112/bert_gnn_arxiv
Jan 2021 - Sep 2021
Contributions:19 commits, 17 pushes, 1 branch 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.