Liam Li
Research Scientist at Pokee AI
United States
Summary
🤩
Rockstar🎓
Top SchoolLiam Li is a research scientist and technical ML leader with a Ph.D. in Machine Learning from Carnegie Mellon and over a decade of experience building production-grade AI systems for startups and enterprises. He helped scale MLOps and distributed training at Determined AI (now HPE), contributed a PyTorch ImageNet NAS integration to the popular determined-ai open-source platform, and has shipped task-specific LLMs at fast-moving startups. His work spans AutoML, distributed training, RAG and generative AI, combining deep research credentials (4k+ citations) with hands-on engineering and product delivery. At Pokee AI he now focuses on AI agents, bringing both academic rigor and practitioner pragmatism to agent architectures and deployment. Notably, he transitioned from quantitative roles in finance to ML research, reflecting a long-standing strength in applied math and real-world systems.
10 years of coding experience
10 years of employment as a software developer
California Institute of TechnologyDoctor of Philosophy - PhD Computer Science - Machine Learning, Doctor of Philosophy - PhD Computer Science - Machine Learning at Carnegie Mellon UniversityStuyvesant High School
Github Skills (10)
computer-vision10
distributed-training10
deep-learning10
nas10
neural-architecture-search10
data-science9
tensorflow9
machine-learning8
pytorch8
hyperparameter-optimization8
Programming languages (6)
Github contributions (5)
determined-ai/determined
May 2020 - Dec 2022
Determined is an open-source machine learning platform that simplifies distributed training, hyperparameter tuning, experiment tracking, and resource management. Works with PyTorch and TensorFlow.
Role in this project:
ML Engineer
ML EngineerContributions:314 reviews, 85 commits, 183 PRs in 2 years 7 months
Contributions summary:Liam implemented a distributed training example for the ImageNet dataset using a state-of-the-art Neural Architecture Search (NAS) architecture, focusing on PyTorch. The core contribution was integrating an Imagenet NAS architecture using PyTorch. The user also added training tricks to the imagenet gaea example.
pytorchdata-sciencedeep-learningdistributed-trainingmachine-learning
determined-ai/lunch_and_learn
Jan 2021 - Jun 2021
Determined's Lunch & Learn Series
Contributions:6 commits, 25 pushes, 1 branch in 4 months
lunch
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.