I am a 2nd-year Ph.D. student in Computer Science at National Yang Ming Chiao Tung University (NYCU), advised by Prof. An-Zi Yen. My research interests center on large language models, information retrieval, question answering, and multimodal assistant.
Before starting my Ph.D., I worked on LLM-driven business intelligence systems at TSMC and language-model classification systems at MediaTek. I am broadly interested in building reliable NLP systems that connect research ideas with practical user-facing workflows.
Routes each query to an appropriate LLM by verbalizing its difficulty, retrieving difficulty-matched examples, and balancing performance against cost.
Introduces a controllable video-synthesis framework for creating realistic egocentric scenarios in which AI agents provide reactive and proactive daily assistance.
Introduces concept-level deficiency prediction and a benchmark for diagnosing the concepts students are likely to struggle with on future problems.
Presents a 3,000-question benchmark with 16 directive types for measuring and mitigating LLM susceptibility to misleading multiple-choice options.
Studies feedback generation for mathematical learning support using authentic student solutions and teacher feedback.
Analyzes how LLMs react when multiple-choice options include malicious or misleading choices.
Compares agentic and decomposed information-retrieval workflows for generating structured financial reports.
Uses agent responses for contrastive and generative pre-training in customer intent detection.
Developed an LLM-driven chatbot for business intelligence use cases, including cross-document summarization and question answering with RAG.
Fine-tuned language models to classify engineer queries automatically.
Built backend and database systems for an alumni communication platform.
Reviewer
Email: tommytyc.cs13@nycu.edu.tw
Website: https://tommytyc.github.io