Associate Professor @ Shenzhen Institutes of Advanced Technology (SIAT), Chinese Academy of Sciences (CAS)
🎓 Ph.D. in AI & Engineering, Curtin University, Australia | B.Eng & M.Eng, Chongqing University

I am an AI System Architect and Researcher operating at the intersection of Agentic Retrieval-Augmented Generation (RAG), Multimodal Large Language Models (LLMs), and Knowledge Graphs (KG). I specialize in designing fault-tolerant, budget-aware autonomous agents capable of structuring highly noisy, real-world data (PDFs, DOCX, images) into actionable intelligence.

If it involves parsing the unparsable, extracting the invisible, or orchestrating multi-agent systems to solve domain-specific engineering challenges—I'm building it.

Core: Python, PyTorch, LangChain, Celery, Llama, DeepSeek, Claude
Databases/Graph: Neo4j, PostgreSQL, Qdrant

• [2025.08] 🥇 1st Place - Alibaba Global LLM Document Parsing Challenge (1/406 globally)
• [2025.11] 🚀 Finalist - AI-Agent Global Track, China AI Application Scenarios Innovation Challenge (Top 8/500)
• [2023.01] 💼 Principal Investigator - National Natural Science Foundation of China (NSFC) Youth Fund (2023-2025)
• [2022.12] ⭐ Academic Rising Star - Shenzhen Institutes of Advanced Technology, CAS
• [2022.11] 🥈 2nd Prize - Hong Kong Construction Industry Council (CIC) Construction Innovation Award

My research focuses on deep learning, ontology, and LLM-driven decision-making in complex environments (e.g., AEC - Architecture, Engineering, and Construction).

Selected Core Papers (First/Corresponding Author):

• [Advanced Engineering Informatics 2025] Retrieval augmented generation-driven information retrieval and question answering in construction management (ESI Highly Cited, Q1)
• [arXiv 2025] AuditAgent: Expert-Guided Multi-Agent Reasoning for Cross-Document Fraudulent Evidence Discovery
• [Advanced Engineering Informatics 2024] A KG-LLM-Enabled Inspection Work Packaging Approach in Module Prefabrication (Q1)
• [Automation in Construction 2024] Safety Consultation for Prefabricated Construction: A Localized Retrieval-Augmented Generative Question-Answering System (Q1)
• [Automation in Construction 2023] Unlocking the Power of Language Models for Smart Configuration in the AEC Domain Using Spoken Language Understanding (Q1)

🔗 For a complete list of publications, citations, and patents, please query my Google Scholar.

threads:
  - id: 0x01
    status: "ACTIVE"
    task: "Ontos-AI/Knowhere (Open Source): A unified Agentic RAG pipeline with dynamic WorkflowOrchestrator and precise hierarchical context rendering."
  - id: 0x02
    status: "ACTIVE"
    task: "Snap-fill Ecosystem (Local): Production-grade document intelligence platform featuring Term-First + Fallback retrieval and Lineage-Based Protection."
  - id: 0x03
    status: "ACTIVE"
    task: "Hierarchy-as-Harness: Researching the use of explicit document hierarchies as an action and evidence scaffold for budget-constrained agents, outperforming Flat-ReAct."
  - id: 0x04
    status: "ACTIVE"
    task: "What and Where (SCD Table Analysis): Investigating spatial cognitive dissonance in LLMs parsing complex tables, demonstrating multi-target interference drives displacement errors."
  - id: 0x05
    status: "ACTIVE"
    task: "WWW 2026 Research (T-Fus / Tree_RAG): Developing incremental fragmented knowledge fusion in RAG via top-down tree structuring."