cv

M.S. in Computer Architecture

B.S. in Computer Science and Technology

Inference Engine Technology Expert

• Responsible for the architecture design and optimization of the MNN inference engine (13K+ GitHub Stars), including operator performance optimization, static memory allocation, and automatic operator fusion.
• Led the development of model conversion tools for TorchScript, Tensorflow, and ONNX, with a focus on supporting control flow for advanced speech models (e.g., ASR, NLU).
• Developed NPU backends (CoreML, NNAPI) to accelerate algorithms like super-resolution and beauty filters in major apps such as Taobao and Taobao Live.
• Initiated and led the MNN-LLM project (1.5K+ GitHub Stars) for on-device deployment of large language models, involving deep optimization of Transformer models (quantization, memory, computation) to support over 100 open-source LLMs.
• Enhanced the MNN ecosystem by developing cv/audio processing modules, refactoring Python interfaces, and building CI/CD automation pipelines, resulting in 2K+ stars for related demo projects.

Compiler Engineer Intern

• Focused on operator optimization for Cambricon chips, compiler development, and linker construction.

MNN-LLM: A Generic Inference Engine for Fast Large Language Model Deployment on Mobile Devices

• MM Asia 2024

Walle: An End-to-End, General-Purpose, and Large-Scale Production System for Device-Cloud Collaborative Machine Learning

• OSDI 2022

IEEE AICAS 2025 Grand Challenge: LLM Software and HardWare System Co-optimization

• the First Prize

IEEE AICAS 2024 Grand Challenge: LLM Software and HardWare System Co-optimization

• the First Prize

Tecorigin Operator Development Competition

• the First Prize

SOPHGO TPU Program Competition

• the First Prize