A problem isn't truly solved until it's solved for all. Googlers build products that help create opportunities for everyone, whether down the street or across the globe. Bring your insight, imagination and a healthy disregard for the impossible. Bring everything that makes you unique. Together, we can build for everyone.\n\nCheck out our career opportunities at goo.gle/3DLEokh

A Family of Matrix Completion Sparse quasi-Newton Method with Super-linear Convergence Property
- Reported on 5th Sino-Japan Optimization Meeting.

Sequential Quadratic Programming Method without Penalty or Filter for General Constrained Nonlinear Programming
- Reported on 21st International Symposium on Mathematical Programming.

Mixed Recommendation Model for User Preference Prediction
- Item-based collaborative filtering and SVD model, programming with C++ and OpenMP.
- Tested with large data provided by Yahoo Music. Won 2nd place in KDD Cup with error rate 2.4808%.
- Paper published on Journal of Machine Learning Research: Workshop and Conference Proceedings 2012.

Non-linear Trust Region Successive Linear Programming for Standard Pooling Problem
- Proved global convergence property.
- The solver was developed with C++ and Coin-Clp library.
- Tested with real data.
- The efficiency and accuracy are competitive to the famous commercial software Aspen PIMS.

University of Chinese Academy of Sciences

President of ACM Student Chapter of Nankai University 
- In charge of Nankai University Programming Contest and regular algorithm training.
- Maintain Nankai University Online Judge ( http://acm.nankai.edu.cn ).

Nankai University

Senior Staff Software Engineer

Behavior Prediction Infrastructure and Evaluation

Senior Staff Software Engineer & TLM

Staff Software Engineer & TLM

Senior Software Engineer & TL

Software Engineer

Interactive Platform of Scientific Computation
- Develop basic matrix computation algorithms and their parallelization.
- Research and develop Simplex Method for Linear Programming.
- Research and develop Branch & Bound Method for Mixed Integer Programming.
Parallelization of Neutral Network Algorithm. Tested with search engine related data.