- Beijing, China
- Staff
Conduct cutting-edge research on on-device efficiency, focusing on compression and acceleration (FP4/INT4 QAT, speculative decoding, on-policy distillation), on-device Agent and token routing/device-cloud orchestration algorithms, and low-latency privacy-preserving personalized multimodal model structures and inference mechanisms.
Responsibilities
- Define new problems, design new methods, build prototypes, and establish evaluation benchmarks to form verifiable technical hypotheses and experimental closed loops.
- Analyze algorithm-system co-design space considering CPU/GPU/NPU heterogeneous computing and on-device OS constraints (Windows/Linux/Android).
- Collaborate with engineers to implement research outcomes, producing papers, patents, technical reports, and reusable components.
- Participate in defining research directions and roadmaps, guiding team members, and enhancing the department's technical influence in on-device efficiency.
Requirements
- Master's degree or above in Artificial Intelligence, Computer Science, or Electronic Information Technology related fields
