Hello, I'm Ping!


A PhD Candidate in Innovation, Technology & Operations at Rady School, UC San Diego.

My research focuses on designing AI systems and data science platforms that enhance strategic and complex decision-making. My job market paper studies how data science contests can unintentionally incentivize suboptimal modeling strategies due to data splitting practices. Further project pipelines explore the impact of AI feedback and benchmark reliability in LLM evaluations.

I am currently seeking a postdoc or an assistant professor position. If you happen to know any opportunity or would like to chat about anything else, feel free to reach out!

Ping's Profile

Research



My research interest lies in the intersections of GenAI, Data Science, Business Innovation, and Technology Management. The primary methodologies are machine Learning, modeling and game theory, and lab experiments, which are applied in the following working papers.


1. Designing Data Science Contests: The Role of Training vs Test Splits
• Job market paper, joint work with Zhe Zhang and Sanjiv Erat
• Under revision for second-round review at Management Science
• First-prize winner of 2025 POMS PITM Best Student Paper Competition
• Nominee for 2025 Artificial Intelligence in Management (AIM) Best PhD Paper Award
• SSRN Top Downloads on Innovation & Operations (full paper link)
Abstract: Companies organize data science contests to source innovative machine learning solutions for business operations. The current study formulates a model of data science contests to investigate how participants choose their modeling approaches and the incentives they face in the competition, and reveals how data splitting practices can unintentionally incentivize suboptimal modeling strategies.



2. AI Supervisors and Human Creativity
• Joint work with Sanjiv Erat
• To be presented at 2025 INFORMS Annual Meeting
Abstract: With AI-based tools improving at an exponential pace, the day is perhaps not far off when management of knowledge work and of knowledge workers becomes just another skill that can be performed by AI supervisors. The current study investigates how a person’s creative performance is affected by the identity of the feedback giver and the nature of the feedback.



3. Chatbot Arena and LLM Benchmarks
Abstract: Elo-style leaderboard has been widely used in evaluating well-known LLM models such as GPT, Gemini and Claude. However, it inherently bears issues of subjective biases and imbalances. The current study assesses the inefficiency of this type of evaluation from multiple dimensions and explores more reliable approaches in LLM benchmarking.

Teaching



1. Instructor
University of California San Diego, July 2024 – August 2024
• 2025 UC San Diego Academic Senate Excellent Teaching Award
• Course title: Business Project Management (syllabus link), 100% student rating
• Content: Managing projects and related business dynamics with AI and trending software based on real-world applications
• Software: ChatGPT, Radiant, Microsoft Excel Gantt Chart



2. Teaching Assistant
University of California San Diego, September 2020 – Present

• AI-Assisted Customer Analytics (Core of Master’s Program)
○ Content: Applying machine learning to collect, analyze, and act on customer data and create value for both customers and firms
○ Software: ChatGPT, Python, R, Radiant, Docker

• Business Analytics (Core of Master’s Program)
○ Content: Making good decisions in complex business problems with statistical and quantitative models such as decision analysis, regression analysis, optimization and simulation
○ Software: Python, R, Radiant

• Operations, Information Systems and Data Analysis (Core of Master’s Program)
○ Content: Synthesizing information and applying operational metrics for systematic design, business execution, and improvement of operations and partner relationships
○ Platform: Littlefield Simulation

• Supply Chain Analytics (Elective of Master’s Program)
○ Content: Understanding and managing the flows of materials and information in a supply chain
○ Topic: Newsvendor, Inventory Control, Demand Forecasting, Revenue Management

Connect with me