Email: chizhang20041101@gmail.com
Research: Agentic llm | 3D Vision | Large-scale AI Systems | RL
Researcher focused on 3D Vision and Foundation Models. Driven by the pursuit of AGI, with broad interests in interdisciplinary areas:
- Systems: vLLM-style inference frameworks & KV cache optimization.
- Agents: "Think–Evaluate–Act" reasoning loops & tool-use.
- Graphics: Neural rendering (3DGS) & Diffusion models.
Core Contributor | ICLR 2026
- Proposed a frequency-aware training scheduler for large-scale 3DGS.
- Optimized densification: +0.9 dB PSNR & 1.5× training speedup.
Project Lead | CCF-C (Under Review)
- Introduced a reasoning loop to enhance robustness under occlusion without extra backbone costs.
ACM MM 2026 (Under Review) / Research Agent
- Built LLM-based agent pipelines for multi-step reasoning and retrieval.
- Languages: Python, C++/CUDA (GPU Programming), Java.
- Frameworks: PyTorch, Transformers, vLLM, DeepSpeed.
- Core ML: Neural Rendering (3DGS), Diffusion Models, Model Compression (Quantization/Pruning).
- ICPC Asia Regional Contest – Medalist
- RoboCup Robotics Competition – National Second Prize
- IEEE Xtreme – Global Award
- Computer Game Competition – National 3rd Place
- CCF National Programming Contest – Top Ranking
- LLM/MLLM & Vision-Language-Action (VLA)
- AI Agents & High-throughput Inference Frameworks
- 3D Multimodal Learning