I am an algorithm engineer with experience in search advertising and credit risk. My interests include agents, large language models, search, recommendation, advertising, causal inference, internet finance, and credit algorithms.
Outside work, I enjoy travel, photography, and badminton.
Research & related work
A Unified Framework for Modeling Heterogeneous Financial Data via Dual-Granularity Prompting
FinLangNet combines DeepFM with a dual-prompt sequence encoder to model heterogeneous financial records and predict credit risk across multiple future horizons.
Multi-Treatment-DML: Causal Estimation for Multi-Dimensional Continuous Treatments with Monotonicity Constraints in Personal Loan Risk Optimization
A framework for estimating responses to continuous credit offers using residualization, disentangled representations, and treatment sensitivities. This reading examines its assumptions, reported evidence, and the distinction between causal estimation and policy optimization.
ZiGong 1.0: A Large Language Model for Financial Credit
ZiGong adapts Mistral 7B to financial credit tasks through multi-task instruction tuning and TracSeq, a sample influence method that incorporates temporal decay.
Real-Time Tracking With Stabilized Frame
S-Siam adds frame stabilization to Siamese trackers, reducing failures caused by camera shake without retraining the tracking network.
Efficient Yolo: A Lightweight Model For Embedded Deep Learning Object Detection
Efficient YOLO compresses YOLOv3 through sparse initialization and structured pruning for embedded road-object detection. This overview is based on the public abstract.
Adaptive Visual Interaction Based Multi-Target Future State Prediction For Autonomous Driving Vehicles
A two-stream visual interaction model combines appearance and motion cues to predict future states of a changing set of objects in driving videos.
Experience
- Aug 2023–present
DiDi
International Fintech · Credit Risk Algorithms
My work in 2023–2025 included behavioral sequence modeling, graph learning, and deep learning systems for credit risk, alongside research on financial language models.
- 2021–2023
JD.com
Search Advertising Algorithms
Worked on retrieval for search advertising, including heterogeneous graph methods, multimodal retrieval, and retrieval pipelines.
- 2020
ByteDance AI Lab
Algorithm Intern
Worked on 3D human pose and gaze estimation.
Education
- 2018–2021
Beijing University of Posts and Telecommunications · Master’s degree
Artificial Intelligence
Multimedia Communication and Pattern Recognition Laboratory (MCPRL)
- 2014–2018
Beijing University of Posts and Telecommunications · Bachelor’s degree
Communication Engineering
Ranked in the top 5% of my cohort; recommended for admission to a master’s program without the entrance examination.