Search GenAI- Focus on multimodal modeling.Past: LLM OptimizationTPU Performance - Apply quantization, sparsity on Gemini and downstream models to improve finetuning and serving efficiency.

Natural Language Processing and Data Mining• Conduct research on weakly-supervised aspect extraction, sentiment analysis, and opinion summarization. Advised by Dr.Jingbo Shang.• One paper accepted by SDM 2021.

Search Platforms

Built document topic models

Natural Language Processing

Deep Reinforcement Learning• Conducted research on deep reinforcement learning algorithms. Advised by Dr.Sicun Gao.• Implemented several reinforcement learning algorithms such as Policy Gradient, Deep Q Learning, Advantage Actor-Critic, Deep Deterministic Policy Gradient, Proximal Policy Optimization using PyTorch to solve OpenAI Gym environments.• Implemented discrete action space and continuous action space versions. Solved OpenAI control problem Pendulum within 1000 episodes.• GitHub link: https://github.com/peiranli/RL

• CSE100 - Advanced Data Structure• CSE134B - Web Client Languages• CSE136 - Enterprise-class Web Application

• Worked at Reinagel Lab UCSD under the Division of Biological Science of University of California, San Diego. • Developed an Android app using Firebase as a backend service for research simulation.• Programmed a script using Node.js to retrieve the app data and parse to JSON format for data analysis.