Undergraduate, Guangzhou University · PhD Applicant (Fall 2027)
Reinforcement Learning · Embodied AI · Efficient Model Deployment
I am a third-year undergraduate majoring in Network Engineering at Guangzhou University, with a focus on AI systems. My research lies at the intersection of reinforcement learning, embodied AI, and efficient model deployment — closing the gap between RL research and real-world robot and edge hardware.
I am particularly interested in how RL policies can be efficiently deployed on resource-constrained devices (NVIDIA Jetson, edge GPUs, CPUs) through model compression, quantization, and cross-platform optimization. I also work on AI safety and adversarial robustness for learning systems.
A cross-platform RL model deployment and performance benchmarking toolkit. Export Stable-Baselines3 policies to ONNX/TorchScript, build TensorRT engines with FP16/INT8 quantization, and benchmark latency, throughput, and accuracy across x86 GPU, NVIDIA Jetson, and CPU — all from one config-driven CLI with auto-generated HTML reports.
Research value: Provides a reproducible benchmark for studying the accuracy-latency tradeoff of RL policy quantization across hardware platforms, enabling researchers to make informed deployment decisions.
I actively contribute to RL and AI infrastructure projects, focusing on documentation quality and codebase alignment:
RL & Learning: PyTorch, Stable-Baselines3, Gymnasium, MuJoCo
Deployment: ONNX, ONNX Runtime, TensorRT, TorchScript, INT8/FP16 quantization
Edge & Robotics: NVIDIA Jetson (Xavier/Orin), ROS, CUDA, cuDNN
Systems: Python, C++, Linux, Docker, Git, CI/CD
Security: AI safety, adversarial robustness, multimedia security
Download CV (PDF) — available upon request.
Email: ldz@e.gzhu.edu.cn
GitHub: github.com/dafahaha
Location: Guangzhou, China