Projects & Open-Source

Projects & Open-Source

Reinforcement Learning & Systems Intelligence (personal projects)

Overview

I develop algorithms, systems, and tools for reinforcement learning and real-world decision-making. Many projects are released as open-source at Systems-Intelligent-Lab @ GitHub.

Highlighted Projects

PA-MoE — Plasticity-Aware Mixture of Experts for Learning Under QoE Shifts in Adaptive Video Streaming (code). Official implementation of our IEEE Transactions on Multimedia paper, focusing on plasticity and stability under shifting objectives.

Plasticity-Scan — Visualizing Neural Network Plasticity Using a Shifting Regression Task (code). A toolkit to study and visualize how neural networks adapt when the target distribution changes over time.

nettrace — A unified data loader for network bandwidth traces and video chunk metadata (code). Designed for research on adaptive streaming, QoE optimization, and RL-based bitrate adaptation.

Planned Directions

In the future, I am interested in pushing projects in directions such as:

• Reinforcement learning algorithms and theory
• Large-scale and data-efficient RL
• Stable and robust control in real systems
• Continual / sustainable learning and world models

Back to Home.