About Me
- I am currently a Staff AI Engineer at LinkedIn Core AI. I focus on agentic systems, LLM post-training, inference efficiency, and generative recommendations.
- Previously, I was a Research Scientist at Meta.
- I received my PhD in Statistics from Cornell University under the supervision of Giles Hooker, working on interpretable machine learning and uncertainty quantification.
- Before Cornell, I received my BS in Probability and Statistics from Peking University.
Preprints and Publications
Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-Training
Findings of the Conference on Empirical Methods in Natural Language Processing (EMNLP 2026), to appear
Robust Batch-Level Query Routing for Large Language Models under Cost and Capacity Constraints
ACM Conference on AI and Agentic Systems (CAIS 2026)
PACED: Distillation at the Frontier of Student Competence
arXiv preprint arXiv:2603.11178, 2026
Not All Tokens Are Needed (NAT): Token Efficient Reinforcement Learning
arXiv preprint arXiv:2603.06619, 2026
On-Policy Self-Distillation for Reasoning Compression
arXiv preprint arXiv:2603.05433, 2026
Overconfident Errors Need Stronger Correction: Asymmetric Confidence Penalties for Reinforcement Learning
arXiv preprint arXiv:2602.21420, 2026
Semantic Search At LinkedIn
35th ACM International Conference on Information and Knowledge Management (CIKM 2026), to appear
Scaling Up Efficient Small Language Models Serving and Deployment for Semantic Job Search
MLSys, 2026
Approximation Trees: Statistical Reproducibility in Model Distillation
Data Mining and Knowledge Discovery, 38(5):3308–3346, 2024
Analyzing Spatial Heterogeneity of Ridesourcing Usage Determinants Using Explainable Machine Learning
Journal of Transport Geography, 114:103782, 2024
S-LIME: Stabilized-LIME for Model Explanation
Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2021
Unbiased Measurement of Feature Importance in Tree-Based Methods
ACM Transactions on Knowledge Discovery from Data (TKDD), 15(2):1–21, 2021
V-Statistics and Variance Estimation
Journal of Machine Learning Research (JMLR), 2021
SILR: A New Exact Test for Demonstrating That an Effect Exists in Binary Trials
OSF Preprints, 2021
Distilling Black-Box Travel Mode Choice Model for Behavioral Interpretation
Transportation Research Board 99th Annual Meeting, 2020
Service
- Reviewer, International Conference on Learning Representations (ICLR) 2022, 2023, 2024
- Reviewer, International Conference on Machine Learning (ICML) 2022
- Reviewer, Neural Information Processing Systems (NeurIPS) 2021
- Reviewer, Annals of Statistics
- Reviewer, Journal of the American Statistical Association
- Reviewer, Machine Learning (Springer)