EDUCATION

• M.S., Applied Data Science

  University of Southern California, Los Angeles, CA, U.S. (Dec 2023)

• B.S., Information Management and Information Systems

  Beijing Foreign Studies University, Beijing, China (Jun 2021)

  • Awards: National Scholarship (top 0.2%), Merit Student, Beijing Outstanding Graduates

RESEARCH INTERESTS

Recommendation Systems, Search Algorithms, Multi-Objective Optimization, Personalization, Large Language Models

PROFESSIONAL EXPERIENCE

Applied Scientist (Recommendation, Search, MTL, LLM) @ Twitch (Sep 2024 - Present)

• Lead large-scale multi-objective optimization for Twitch’s live-stream recommendation systems, defining trade-offs across engagement and retention metrics in production.
• Architect and drive end-to-end ML ranking pipelines, resolving data quality challenges, deploying deep learning models and aligning offline evaluation with online experimentation.
• Design and scale user-segment-aware multi-model architectures, introducing debiasing strategies and calibrated new-user utility functions.
• Lead model evolution through segment-weighted loss, curriculum learning, and multi-task learning (MTL) frameworks such as the Multi-Gate Mixture-of-Experts (MMoE).
• Enhance Twitch’s search experience through LLM-based query understanding and reformulation.
• Collaborate cross-functionally with engineering and data science to de-ambiguate problem definitions, plan execution, and mentor new team members.

Machine Learning Engineer (Generative AI, Search Engineering, RAG) @ SylphAI Inc. (Apr 2024 - Aug 2024)

• Developed a GenAI chatbot search engine (RAG) by embedding texts, indexing, creating the retriever system, ranking and generating candidate-specific answers.
• Built, trained and fine-tuned deep learning classifiers using PyTorch.
• Created high-quality data labels utilizing Large Language Models(Mistral-7B, GPT-4, and Gemini) with few-shot Prompt Engineering and designed an efficient ETL pipeline to manage data on AWS RDS.
• Implemented agent framework from research with function calls and contributed to an open-source LLM library AdalFlow.

Data Scientist Intern @ Adobe Inc. (May 2023 – Aug 2023)

• Spearheaded cancellation analysis and identify cancellation patterns, optimizing the user journey.
• Innovated a product marketing strategy with an estimated 10% conversion rate improvement and an $8.3M annual recurring revenue (ARR) lift by deep diving on 5+ metrics using 5M+ data to support decision-making.
• Built a Logistic Regression and an XGBoost model with feature engineering to analyze the important and significant factors that lead to cancellation on 3M data processed by SQL.