CV
Curriculum Vitae / Resume
Contact Information
| Name | Kuan-Chieh Lo |
| Professional Title | Ph.D. Student in Computer Science & Engineering |
| lo.311@osu.edu | |
| Location | Columbus, Ohio |
Professional Summary
Ph.D. student at The Ohio State University working on LLM & Agentic AI, misinformation detection, fairness in federated learning, and LLM safety.
Education
-
2022 - 2027 Columbus, Ohio
Ph.D.
The Ohio State University
Computer Science and Engineering
- Advisor: Dr. Srinivasan Parthasarathy
- Research: LLM & Agentic AI, Misinformation, Fairness, LLM Safety
-
2012 - 2014 Taiwan
M.S.
National Taipei University of Technology
Electrical Engineering
-
2008 - 2012 Taiwan
B.S.
National Changhua University of Education
Mechatronics Engineering
Research Experience
-
2022 - present Columbus, Ohio
Graduate Student Researcher
The Ohio State University
- Multi-Agent Adversarial Claim Verification — Proposed a multi-agent LLM framework organizing heterogeneous agents across multiple foundation models, incorporating claim decomposition, multi-hop knowledge retrieval, and adversarial verification. (ICWSM 2026)
- Agentic AI for Crisis Response — Developed Crisis Observatory, a multi-agent LLM system for crisis response that extracts credible signals from social media by integrating topic modeling, geolocation extraction, and a RAG pipeline. (ICDM 2025)
- Fairness in Federated Graph Learning — Proposed FairWAG, a fairness-aware federated learning framework that applies Shapley Values to quantify client contributions, enabling adaptive aggregation weights. (EAAMO 2025)
- LLM Safety: Jailbreak Attacks on Large Reasoning Models — Uncovered a prompt injection vulnerability in Large Reasoning Models (LRMs), where adversarially crafted prompts inject a spoofed chain-of-thought block to bypass safety alignment. (Under Review)
- Compositionality Evaluation of VLMs — Assessed compositional reasoning capabilities of state-of-the-art vision-language models (VLMs), exposing critical skill gaps across object detection, relational extraction, and attribute binding tasks.
-
2019 - 2022 Taipei, Taiwan
NLP Research Scientist
Academia Sinica
Mentor: Dr. Lun-Wei Ku
- Misinformation Mitigation via Recommendation Systems — Developed VICTOR, a reinforcement learning-based module that implicitly re-ranks news recommendations to surface verified articles. (WWW 2022, WSDM 2021)
- Echo Chamber Reduction — Built a news-analysis platform that applies a stance classification model to present multi-source perspectives on events, reducing filter bubble effects. (WWW 2021)
- Visual Storytelling Evaluation — Developed Vrank, a reference-free automatic evaluation metric for visual storytelling (VIST), achieving ~30% higher accuracy than existing metrics. (ACL 2022)
Work Experience
-
2014 - 2018 Taiwan
Software and Firmware Engineer
ASUSTeK Computer Inc.
- Developed system firmware for ASUS products, including gaming desktops, business and consumer laptops.
Skills
Academic Services
-
- Reviewer
Conference Reviewer
- ICML (2026), ICWSM (2026), NeurIPS (2024), ACL (2024, 2022), AAAI (2022), IAAI (2022, 2021, 2020), NLPCC (2021), EMNLP (2020)
Teaching
-
2023 - present Columbus, Ohio
Graduate Teaching Associate
The Ohio State University
- Database Systems (Autumn 2023, Spring 2024)
- Data Mining (Autumn 2024, Autumn 2025, Spring 2026)
- Network Science (Spring 2025)
Projects
-
Creativity in Large Language and Vision-Language Models
Investigated the compositional factors influencing creative output in LLMs and vision-language models (VLMs), analyzing how individual components (e.g., concept, style, structure) interact to drive creativity.
-
LLM-Powered Agentic Chatbot for Bridal Consultation
Designed and developed a multi-agent AI system for a bridal company integrating a conversational LLM chatbot, a personalized recommender system agent, and a meeting scheduling agent into a unified pipeline.
-
LLM-Assisted Legal Document Processing System for Law Firms
Built an NLP pipeline for law firms leveraging LLMs and retrieval-augmented generation (RAG) to automatically retrieve semantically relevant past legal cases and populate structured legal documents.