CV
Full academic CV is available as a PDF (button to the right) or webpage (provided below).
Contact Information
| Name | George Pu |
| Professional Title | Applied Scientist |
| pu.george99@gmail.com | |
| Location | Seattle, WA |
| Website | https://georgerpu.github.io |
Professional Summary
Applied Scientist at Amazon building AI agents and ML systems at scale. Published researcher in deep learning and federated learning. 4+ years of experience spanning research, ML infrastructure, and software development.
Experience
-
2025 - Present Seattle, WA
Applied Scientist
Amazon, Sustainability Science and Innovation
- Developed deep research agent that auto-generates bills of materials, extending actionable environmental impact analysis to 75% of Amazon’s carbon footprint with a <30% MAPE.
- Led sustainability scientists to annotate hundreds of life cycle inventories/assessments, creating 2 new internal benchmarks.
-
2024 - 2025 Seattle, WA
Machine Learning Engineer / Applied Scientist
Amazon, Customer Trust
- Created time series classification models that helped reduce bad seller debt by an estimated $3 million.
- Trained LightGBM models to reduce investigator case load, saving an estimated $500,000 per year.
- Trained graph machine learning models on SageMaker.
-
2021 - 2024 Seattle, WA
Software Developer/Engineer
Amazon, Customer Trust
- Automated bad actor investigation case creation, saving 2 hours of investigator time per week.
- Onboarded ML models to SageMaker, saving >184 hours annually of scientist effort retraining models and batch inferencing.
- Designed and implemented an offline graph query execution engine over hundreds of millions of nodes on Amazon Neptune that helped stop $500,000+ in fraudulent sales; work featured on the AWS Database Blog.
-
2017 - 2021 Gainesville, FL
Research Assistant
University of Florida
- Collaborated with a PhD student and professor to fit multi-compartment epidemiological models to Leishmaniasis data.
- Published 5 papers at top venues (IROS, NeurIPS) with over 250+ citations.
Education
-
2017 - 2021 Gainesville, FL
Publications
-
2021 Seeing Through Walls: Real-Time Digital Twin Modeling of Indoor Spaces
2021 Winter Simulation Conference
George Pu, Paul Wei, Amanda Aribe, James Boultinghouse, Nhi Dinh, Fang Xu, Jing Du.
-
2020 Communication-Efficient Federated Learning via Dataset Distillation
2020 NeurIPS Workshop on Scalability, Privacy, and Security in Federated Learning (SpicyFL)
Yanlin Zhou*, George Pu*, Xiyao Ma, Xiaolin Li, Dapeng Wu. (* equal contribution)
-
2019 Adaptive Leader-Follower Formation Control and Obstacle Avoidance via Deep Reinforcement Learning
2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
Yanlin Zhou*, Fan Lu*, George Pu*, Xiyao Ma, Runhan Sun, Hsi-Yuan Chen, Xiaolin Li. (* equal contribution)
Skills
Languages: Java, Javascript, Python, SQL, Typescript
Python Packages: AutoGluon, Langchain, Matplotlib, Numpy, Pandas, Polars, PyTorch, Scikit-learn, Tensorflow
AI/ML: Agents, LLM Evaluation, Deep Learning, Graph Neural Networks, Prompt Engineering, RAG, Tabular Machine Learning, Time Series Classification
Cloud: AWS, Bedrock, CloudWatch, DynamoDB, IAM, Neptune, S3, SageMaker, Step Functions