Charles Sun - Waabi | LinkedIn (original) (raw)
About
I work on deep reinforcement learning and NLP as part of the Robotic AI & Learning Lab at…
Experience & Education
Waabi
View Charles’s full experience
See their title, tenure and more.
Projects
Dec 2021
Offline RL framework in JAX (Flax) that supports
- Distributed multi-device training (GPU/TPU)
- Multiprocessed dataloaders (similar to PyTorch)
- Modular experiment management and logging
Currently used by myself for research.
See project
Aug 2019 - Feb 2020
• Implementing various reinforcement learning algorithms in Python with TensorFlow 2.0 and PyTorch, tested in OpenAI Gym environments.
• Open-source, environment-agnostic, completely general parametrized implementation giving users total control over settings and environments without changing code.
• Implemented Q-Learning, Policy Gradient, Actor-Critic, GAE, PPO, DDPG, TD3
See project
Mar 2017 - Aug 2018
• Created procedurally generated world represented by cubes (voxels) using UE4 and C++.
• Designed custom world generation algorithm utilizing assorted noise algorithms.
• Supported near-infinite world creation by utilizing compression techniques.
• Solved concurrency issues with data chunking due to the need to sync up multiple clients.
• Difficult problem due to a dynamic changing world creating the need to send large amount of data between clients with minimal latency.
See project
-
Deferred rendering engine created using modern OpenGL. Greatly improves performances over traditional forward rendering engines, allowing for a massive amount of dynamic lighting without much performance loss. Includes framework for adding extra post-processing effects such as bloom, ambient occlusion, FXAA, and HDR.
See project
Languages
Chinese
Native or bilingual proficiency
English
Native or bilingual proficiency
Other similar profiles
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.