Irvine, CA *****
Cell: 949-***-**** (primary)
Seeking a position as a researcher and modeler using my background and experience in applied mathematics, probability, statistics and machine learning. Skills
SAS Enterprise Guide, SAS/Base, SAS/STAT, SAS/ETS, SAS/OR, SAS/Oracle, SAS/IML, SAS/Access, SQL, Transact-SQL, C\C++, R, Python, Matlab, Microsoft Office (Word, Excel, PowerPoint, Access), Microsoft SQL server Database, Credit risk management, Financial portfolio performance forecast, PD/LGD/EAD technique and Vintage Analysis, Statistical modeling and machine learning, Monte Carlo simulation, Survival Analysis, Logistic Regression, Linear Regression, Generalized Linear Models, Moody’s Analytics, Haver Analytics Summary
I was a math teacher in a university in China before 1995. After I came to the US and finished my Ph. D study, I started my career in 2000. In the last 18 years, I mainly worked in three companies, Veros Software, UnitedHealth Group, Hyundai Capital America and Veros Credit (Auto finance). My main work was closely related to mathematical and statistical analysis and modeling. Work History
1. November, 2018 – Present: Senior Data Scientist, Veros Credit, Irvine, California.
This is a subprime auto financing company. My job is to provide data and modeling support for the financing business, including dealer service, loss forecast, data analysis, etc.. Statistic Analysis and Machine Learning are the main tools I have been using, including but not limited to time series
(ARIMA), decision tree, linear and logistic regression, survival analysis, etc.. The software I use are mainly Python, SAS and Microsoft Sql Server
2. July, 2015 – August, 2018: Statistician, Hyundai Capital America, Irvine, California.
This is an auto financing (auto lending) company. I have been working in the risk management area to build models assessing and evaluating the risk of portfolios, and planning business strategy. Here are some specific items I have worked on:
1) Prepayment model. I worked on this model on vintage level (cohort level), and each vintage consists of those accounts that were originated in a specific time period and may have some other natures. The model goal was to forecast prepayment percentage of any vintage anytime in the future.
2) Termination model. Termination is defined as any action that makes an active account become inactive. Those actions include paid-off, charge- off, grounding, repossession, etc. The model output is a probability distribution in the lifetime for each account.
3) Credit Loss Model at vintage level. This mode was built on vintage level. The goal is to forecast credit loss in the future. This is full set of loss models that includes active balance and unit, charge-off balance and units, recovery balance and delinquency balance and units. 4) Recovery model. This model is different from the model included in 3) as this is at account level. The purpose of this model is to assist collection strategy. The output includes, for each charge off account, the maximum possible recovery amount, and percentage of the recovery for each following month.
5) Expected Credit Loss Model at account level with PD/LGD/EAD methodology. This model is compliant to CECL(current expected credit loss) standard, among other thing, producing expected loss at account level in lifespan.
6) The data that I have been using includes internal loan specific data, credit bureau data (like from Experian) and Moody’s macroeconomic data. 3. July, 2012 – July, 2015: Senior Biostatistician, OptumRx, United Health Group, Irvine, California.
Here are a few of main areas I had been working on: 1) Quarterly intervention programs including “Disease Modifying Anti- Rheumatic Drug Use in Patients with Rheumatoid Arthritis”, “ACEI/ARB Use in Patients with Diabetes and Hypertension”, etc.. 2) PDC (proportion of days covered) indicator which is one of the indicators that we frequently use to measure how adherent patients stick to the prescribed medications.
3) Propensity score matching which is an effective way to match different groups of patients based on a single indicator, instead of multiple characteristics.
4) Sample size (patients) estimation which is a crucial step before a drug effect comparison for multiple groups of patients can be carried out. 4. April, 2001 – July, 2012: Senior Research Scientist, Veros Software Inc., Santa Ana, California.
I had been working on real estate appraisal and the capital management: 1) Automated Validation Model (AVM). This project is to forecast the home prices based on the historical sales and home related characteristics like square footage, lot size, bedroom number, etc..
2) Capital Management. This project was to analyze the stock prices based historical values.
5. Nov, 2000 – Feb, 2001: Engineer, Unipass Technologies (this company later broke into two, one of which becomes Veros Software Inc.), Irvine, California.
I had been working on the same projects as stated in Veros Software Inc. 6. July, 1999 – June, 2003: Postdoctoral Research Fellow, Claremont Graduate University, Claremont, California.
I was working as a researcher, algorithm developer and computer programmer in the projects sponsored by various companies and institutions (such as Los Alamos National Laboratory, ADAC Laboratory, University of California at Irvine, etc).
1) Neutron transport problems. My job was to construct and analyze models using statistics, differential equations and some necessary physics and to develop various kinds of accelerated Monte Carlo algorithms, sequential correlated sampling and adaptive importance sampling methods, to obtain global solutions
2) Electron transport problems. The goals were to characterize tissue optical properties and to plan radiation treatments.
7. May, 1996 – August, 1996: Graduate Research Assistant, Los Alamos National Laboratory, Los Alamos, New Mexico.
My job was to construct and analyze models, test some new ideas in Monte Carlo methods and assist staff members to prepare presentations. 8. November, 1986 – September, 1994: Assistant Professor, Changsha Institute of Technology, Changsha, China.
1) Research in probability and statistics, differential equations. 2) Computer programming in FORTRAN, PASCAL, C and, later, in C++ 3) Teach undergraduate and graduate students advanced mathematics, such as, calculus, probability, statistics, numerical analysis, differential equation, topology, differential geometry.
1. Ph. D, Department of Mathematics, Claremont Graduate University, Claremont, California, September, 1995 – May, 1999. 2. M.S., Wuhan Institute of Mathematical Sciences of Academia Sinica, Wuhan, China, February, 1982 – January, 1985.
3. B.S., Northwestern Polytechnic University, Xi’an, China, February, 1978 – December, 1982.
1. R. Kong, “The Sufficient Condition for the Cauchy Problems of a Class of PDE’s with Triple Characteristics”, Acta Mathematica Scientia, 6(1986), 1, 109-119;
2. R. Kong, “The Singular Initial Value Problem for a Class of PDE’s”, Journal of National University of Defense Technology, No. 3, 1988; 3. R. Kong, “The Existence and Uniqueness of the Global Classical Solution to the Cauchy Problems of Some Semilinear Heat Transfer Equations”, Journal of National University of Defense Technology, No. 3, 1992; 4. R. Kong & J. Spanier, “Sequential correlated sampling algorithms for transport problems”, Monte Carlo and Quasi-Monte Carlo Methods 1998, Springer-Verlag, New York, H. Niederreiter and J. Spanier, Eds; 5. R. Kong & J. Spanier, “Error analysis of sequential Monte Carlo methods for transport problems”, Monte Carlo and Quasi-Monte Carlo Methods 1998, Springer-Verlag, New York, H. Niederreiter and J. Spanier, Eds. 6. R. Kong & J. Spanier, “Residual Versus Error in Transport Problems", Monte- Carlo and Quasi Monte-Carlo Methods 2000: Proceedings of a Conference at Hong Kong Baptist University, eds.K.-T. Fang, F.J. Hickernell, and H. Niederreiter, 2002.
7. J. Spanier and R. Kong, “A New Adaptive Method for Geometric Convergence”, Monte-Carlo and Quasi Monte-Carlo Methods 2004: Proceedings of a Conference in Singapore, Harald Niederreiter, ed., Springer- Verlag, Berlin, 2004, pp. 439-449, 2004.
8. R. Kong, A. Martin & J. Spanier, “Efficient, Automated Monte Carlo Methods for Radiation Transport”, J. Comp. Phys., vol. 227, pp. 9463-9476, 2008.
9. R. Kong & J. Spanier, “A New Proof of Geometric Convergence for General Transport Problems Based on Sequential Correlated Sampling Methods”, J. Comp. Phys., vol. 227, pp. 9762-9777, 2008.
10. K. Bhan, R. Kong and J. Spanier, "Adaptive Monte Carlo Algorithms Applied to Heterogeneous Transport Problems", Monte Carlo and Quasi-Monte Carlo Methods 2008, pp.209- 225, Springer-Verlag, 2009.
11. R. Kong and J. Spanier, "Geometric convergence of second generation adaptive Monte Carlo algorithms for general transport problems based on correlated sampling", Int. J. Pure Appl. Math., 59, 435-455, (2010). 12. R. Kong and J. Spanier, "Geometric Convergence of Adaptive Monte Carlo Algorithms for Radiative Transport Problems Based on Importance Sampling Methods", NUCLEAR SCIENCE AND ENGINEERING: 168, 197–225
13. R. Kong and J. Spanier. "Transport-constrained extensions of collision and track length estimators for solutions of radiative transport problems." J. of Comp. Phys. 242 (2013) 682-695.
14. R. Kong and J. Spanier. “A New Proof of Geometric Convergence for the Adaptive Generalized Weighted Analog Sampling (GWAS) Method”. Monte Carlo Methods and Applications. ISSN (Online) 1569-3961, ISSN (Print) 0929-9629, DOI: 10.1515/mcma-2016-0110, June 2016. 15. S. Zhao, R. Kong and J. Spanier. “Towards Real-Time Monte Carlo for Biomedicine”, MCQMC 2016: Monte Carlo and Quasi-Monte Carlo Methods pp 447-46.