Third-year undergraduate at Guangzhou University, Network Engineering. Research focus: reinforcement learning, embodied AI, and efficient model deployment — bridging RL research and real-world edge hardware.
Applying for PhD programs (Fall 2027) in Embodied AI / Reinforcement Learning / Robotics.
Academic Homepage | CV | ldz@e.gzhu.edu.cn
Interests:
- Reinforcement learning for embodied agents (off-policy algorithms, sample efficiency)
- Cross-platform RL model deployment (GPU / Jetson / CPU)
- Model compression and quantization for RL policies (INT8 / FP16)
- AI safety and adversarial robustness
Featured Project — rl-deploy-bench
Cross-platform RL model deployment and benchmarking toolkit. Export SB3 policies to ONNX/TorchScript, build TensorRT engines with FP16/INT8 quantization, and benchmark latency/throughput/accuracy across x86 GPU, Jetson, and CPU — from one config-driven CLI with auto-generated HTML reports.
Merged
| Project | PR | Contribution |
|---|---|---|
| shmuma/ptan | #57 | Fixed Gymnasium API compatibility in experience sources |
| Algorineko/AgenticArXiv-RL | #72 | Added MIT License, CONTRIBUTING, issue templates, CI |
| NVlabs/FluxVLA | #122, #123 | Fixed typos and installation docs |
In Review — Code Bug Fixes
| Project | PR | Contribution |
|---|---|---|
| scikit-learn/scikit-learn | #34986 | Removed int(c) coercion in compute_class_weight that broke dict lookup for string labels parsing as int (Fixes #34883) |
| DLR-RM/stable-baselines3 | #2289 | Fixed DummyVecEnv/SubprocVecEnv ignoring reset options/seeds on auto-reset |
| microsoft/TextWorld | #377 | Fixed env.step crash on multi-command input causing int() parse failure |
| google/brax | #676 | Fixed EpisodeWrapper metrics accumulation with action_repeat |
| huggingface/datasets | #8633 | Fixed pathlib.Path regression in dataset loading |
| huggingface/diffusers | #14796 | Code improvement |
In Review — Docs / Other
| Project | PR | Contribution |
|---|---|---|
| UoA-CARES/cares_reinforcement_learning | #409 | Algorithm docs rewritten against actual codebase API |
| OWASP/secure-agent-playbook | #28 | Fixed outdated OWASP LLM Top 10 IDs |
RL — PyTorch · Stable-Baselines3 · Gymnasium · MuJoCo Deployment — ONNX · ONNX Runtime · TensorRT · TorchScript · INT8/FP16 Edge/Robotics — NVIDIA Jetson · ROS · CUDA · cuDNN Systems — Python · C++ · Linux · Docker · Git · CI/CD