Haoyang Liu

◦ Based on the "Qwen" large model platform of China Merchants Group, my team was responsible for designing and developing large model application solutions for vertical industries (ship repair and construction), while I was responsible for building the enterprise knowledge engineering system, including the cleaning, structured processing and vectorized storage ofhistorical documents/data, and thus providing a high-quality knowledge base for LLM. ◦ I also participated in developing and integrating intelligent agents (Agents) and using LLM API to realize natural language interaction and automatic processing of complex tasks (such as intelligent parsing of test reports, structured information extraction and automatic storage). ◦ Through the task, I got proficient in mainstream large model API calls (Python), and applied fine-tuning, enhanced retrievalgeneration (RAG) and other technologies to optimize model performance and task adaptability. ◦ Core achievements: I lead the intelligent report generation module in the "Automatic Generation of Hull Structure Fatigue Analysis Report" project and realize the structured extraction and automatic generation of report contentthrough LLM drive, reducing the average time of generating a report from six hours to twenty minutes.

Identified 10+ opportunities through searching for exhibitions and collaborators, leading to a 25% increase in online engagement. Expanded sales channels by partnering with up to 8 corporations, boosting the Canton region’s market coverage by 30%.

Researched Hurwitz Class numbers under Prof. Ben Kane, focusing on error analysis and computational modelling. Derived error bounds for the First moment and the exact form of the Second-moment Hurwitz Class Number based on Python and LMFDB. Written a report paper on these contributions and presented findings at departmental seminars.

Proof Verification & Reconstruction: Identified ambiguities in 3 critical proofs in quantum computing papers. Co-developed numerical validation frameworks (Python/Matlab), and successfully filtered through more than 50 papers, efficiently distinguishing relevant research from unrelated work.

Construct an analysis of the petroleum price based on the VAR and Arma model, with all calculations and diagram constructions completed through R. Finished a report paper based on the work done.

GPA: 3.4 / 4.3 (Higher than B+) Honors: HKU Entrance Scholarship (Fall, 2021) and HKU Short-term Reach-out Award (Summer, 2023) Relevant Coursework: Machine Learning, Linear Algebra, Calculus, Statistics, Probability Theory, Optimisation, Discrete Mathematics, Numerical Analysis, Scientific Computing, Analysis, Accounting, Microeconomics, Data Structure, Algorithm