Oct 2023 – Jun 2025 · 1 yr 8 mos · Greater Seattle Area

• TL for Search and Recommendation in Central Machine Learning team.
• 𝐎𝐩𝐞𝐧 𝐒𝐨𝐮𝐫𝐜𝐞 𝐏𝐫𝐨𝐣𝐞𝐜𝐭𝐬
• https://github.com/amazon-science/MO-LightGBM
• https://github.com/amazon-science/DeepMTL2R
• 𝐒𝐞𝐥𝐞𝐜𝐭𝐞𝐝 𝐏𝐮𝐛𝐥𝐢𝐜𝐚𝐭𝐢𝐨𝐧𝐬
• 𝗟𝗟𝗠
• Generative Prompting for Complex Product Retrieval, TheWebConf 2025
• AutoEval-ToD: Automated Evaluation of Task-oriented Dialog Systems, NAACL 2025
• Q-Tuning: Queue-based prompt tuning for lifelong few-shot language learning, NAACL 2024
• Large Language Models are Privacy-Preserving Re-Rankers for Information Retrieval, 2024
• 𝗦𝗲𝗮𝗿𝗰𝗵
• MO-LightGBM: A Library for Multi-objective Learning to Rank with LightGBM, SIGIR 2025

Feb 2022 – Sep 2023 · 1 yr 7 mos · Greater Seattle Area

• Building and deploying advanced algorithms for Search and Recommendation in Central Machine Learning team.
• 𝐒𝐞𝐥𝐞𝐜𝐭𝐞𝐝 𝐏𝐮𝐛𝐥𝐢𝐜𝐚𝐭𝐢𝐨𝐧𝐬
• 𝗦𝗲𝗮𝗿𝗰𝗵
• Querywise Fair Learning to Rank through Multi-Objective Optimization, KDD 2023
• Multi-Label Learning to Rank through Multi-Objective Optimization, KDD 2023
• 𝗥𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱𝗮𝘁𝗶𝗼𝗻
• G-STO: Sequential Main Shopping Intention Detection via Graph-Regularized Stochastic Transformer, CIKM 2023
• 𝗠𝘂𝗹𝘁𝗶-𝘁𝗮𝘀𝗸 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴
• Federated Multi-Objective Learning, NeurIPS 2023

Jun 2020 – Jan 2022 · 1 yr 7 mos · Greater Seattle Area

• Work on Search and Recommendation in Central Machine Learning team.
• Launched multiple search ranking models that significantly boost sales and product delivery speed while reducing cost-to-serve for Amazon.
• Conducted research on efficient DNN training, multi-task learning, multi-label learning to rank, and product substitute and complementary recommendation. One of our paper was ranked top 10 most viewed paper published in Amazon, 2022.
• 𝐒𝐞𝐥𝐞𝐜𝐭𝐞𝐝 𝐏𝐮𝐛𝐥𝐢𝐜𝐚𝐭𝐢𝐨𝐧𝐬
• 𝗥𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱𝗮𝘁𝗶𝗼𝗻
• Personalized Complementary Product Recommendation, the WebConf 2022
• https://www.amazon.science/latest-news/the-most-viewed-amazon-science-publications-of-2022
• 𝗠𝘂𝗹𝘁𝗶-𝘁𝗮𝘀𝗸 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴
• A multi-task learning framework induced by Pareto stationarity, ICML 2022

Sep 2019 – Dec 2019 · 3 mos · Mountain View, CA

• We develop a principled bilevel model approach for big data subsampling and dramatically reduce the training time of large scale deep learning models while maintaining close to state-of-the-art accuracy, including ImagetNet and Criteo.
• https://arxiv.org/abs/2104.13114

May 2019 – Aug 2019 · 3 mos · Greater Seattle Area

• We build an end-to-end deep learning based recommendation system to jointly learn from customer behavior data and product content information to improve the current product substitute recommendation systems for Amazon customers.