Mesut Yang
SDE II, Stores Foundational AI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolMesut Yang is a software development engineer focused on productionizing AI at Amazon, currently building experiment and data curation platforms for large language models that power shopping experiences. With 11 years of engineering experience across Amazon Search & AI and previous ML integration work at Ironclad, he blends infrastructure-level model evaluation expertise with applied LLM data pipelines. A UC Berkeley MS/BA alumnus, Mesut has deep roots in AI education and research—he co-instructed CS 188 multiple times and authored a thesis on human-AI coordination using deep reinforcement learning. He contributes to cooperative-AI tooling (notably enhancing the Overcooked AI benchmark with CNN-friendly state encodings and random-layout training), showing an appetite for improving benchmarks as well as production systems. Based in San Francisco, he brings a rare combination of teaching, research, and hands-on ML systems engineering to large-scale AI problems.
11 years of coding experience
5 years of employment as a software developer
Master of Science - MS Computer Science, Master of Science - MS Computer Science at University of California, BerkeleyHigh School High School/Secondary Diplomas and Certificates, High School High School/Secondary Diplomas and Certificates at Saint Anthony's High SchoolNon-Matriculated Undergraduate Program Mathematics / Physics, Non-Matriculated Undergraduate Program Mathematics / Physics at Stony Brook University
English, Chinese, Chinese
Github Skills (9)
pytorch10
machine-learning10
deeplearning-ai10
artificial-intelligence10
deep-learning10
reinforcement-learning10
python9
faster-rcnn8
mask-rcnn8
Programming languages (4)
Github contributions (5)
HumanCompatibleAI/overcooked_ai
Jun 2020 - Sep 2021
A benchmark environment for fully cooperative human-AI performance.
Role in this project:
ML Engineer
ML EngineerContributions:21 reviews, 120 commits, 16 PRs in 1 year 3 months
Contributions summary:Mesut contributed to the development of a new state encoding mechanism within the Overcooked AI environment. Their changes focused on featurizing the OvercookedState object into a series of boolean masks that CNNs can easily interpret. These modifications included defining new map layers, and implementing new features like urgency to the state representation. The user improved the environment for multi-layout training by integrating support for randomly generated layouts.
aiartificial-intelligencedeep-learningmachine-learningpytorch
mesutyang97/LibFinder
Apr 2017 - Apr 2017
Contributions:20 PRs, 19 pushes, 18 branches in 5 days
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.