I am an M.S. student in Artificial Intelligence at Shenzhen University, advised by Prof. Dugang Liu. I am currently an LLM Algorithm Intern at MemTensor and previously worked as an NLP Research Intern at Tencent FinTech.
My research focuses on agent memory, post-training for long-horizon LLM agents, and LLM-enhanced recommender systems. I enjoy turning research ideas into reproducible systems, from training and evaluation pipelines to inference optimization and full-stack AI applications.
- HSUGA — First-author work accepted to ACL 2026 Findings. We introduce hierarchical semantic understanding and group-aware alignment for LLM-enhanced sequential recommendation. [Paper]
- Movie Agent Lab — An AI-native movie recommendation platform with conversational agents, memory, RAG, pluggable recommenders, FastAPI, Vue 3, and Docker. Awarded National Top 50 / Innovation Award in the 2023 iFlytek Spark Cup (科大讯飞「星火杯」认知大模型场景创新赛).
- Awesome LLM Tech Reports — Evidence-oriented Chinese deep-reading notes on frontier LLM technical reports, covering architecture, training, post-training, and evaluation.
- RecSys-HR 2026 WorkRB — Open-source solution for the RecSys-HR 2026 WorkRB Challenge, combining embedding retrieval, LLM reranking, and learning-to-rank fusion.
- 2025 China Telecom Tianyi Cloud Xirang Cup (天翼云息壤杯), National Finals, LLM Application Track — 7th nationwide. Team lead. Built a Qwen3-8B mathematical-reasoning pipeline and vLLM-Ascend inference optimization.
- 2023 iFlytek Spark Cup (星火杯) Cognitive Large Model Scenario Innovation Competition — National Top 50 Innovation Award. Project: Movie Agent Lab.
- Core algorithm developer for the RecSys-HR 2026 WorkRB Challenge, achieving an overall nDCG@100 of 0.8809.
- Contributor to SecretFlow SPU: added and improved the Chinese documentation for SGD in MPC-ML. [Merged PR #1055]
- Teaching assistant at Datawhale, supporting open-source courses including Pumpkin Book, CAMEL, Happy-LLM, and practical RAG.
- Former organizer of an AI open-learning community at Taiyuan University of Technology, with 180+ contributors and 50K+ project visits.
- M.S. in Artificial Intelligence, Shenzhen University, 2024–2027
- B.Eng. in Software Engineering, Taiyuan University of Technology, 2019–2023
Agent Memory · LLM Post-training · Reinforcement Learning · RAG · Recommender Systems · Model Evaluation
Python · PyTorch · Transformers · verl · vLLM · SGLang · FastAPI · Docker
- GitHub: lilxmx
- Email: 1142456160@qq.com