Zizhao Hu
AI Researcher · PhD Student at USC · GLAMOUR Lab · MINDS · MOVE Fellow
I build AI systems that improve themselves while remaining under control. My work spans multi-agent systems & synthetic data, brain-inspired neural architectures, and continual learning. Former MOVE Fellow at Handshake AI, specializing in frontier model training and safety.
Make AI Smarter
Training paradigms — self-improving loops, synthetic data generation, continual learning without forgetting
Build AI Systems
Full training pipelines for LLM, VLM — domain-specific, task-oriented AI training and system building
Make AI Fast
Efficient multimodal architectures — low-latency transformers, novel attention, scalable model design
My Vision
Specialized AI — models trained for coding, STEM, and scientific reasoning — will drive civilization forward. Embodied, instruction-tuned AI with emotional intelligence will balance society through human-centered interaction. The bridge between the two: task-specific synthetic data and verified curricula that make training specialized models fast, cheap, and accessible to everyone.