• Payroll report performance tuning: reorganizing complex many table join aggregation into multiple smaller table join aggregation as subquery and then join subquery together.• Payroll support: quick response for database data correction sometimes with java code fix.• Payroll feature development: regulation change such as state W4 form with Junit test.

• Frontend partner tool feature improvement using java script under Polymer framework with Spanner database.• Hotel booking SOAP/REST API bug fix using java8, JSON, YAML Guice, message protocol buffer.• Expedia hotel REST API client side API development with Junit.

• Web based (Jboss) customer service production support: oracle database data fix, plsql, and java bug fix. App Dynamics was used as tool to monitor.• Apache DBCP to DBCP2, Redis Jedis client, and active MQ upgrade and performance evaluation.

• Oracle 11g database tuning: tuning the parameter based on system metric and statistics AWR report for 6TB network surveillance department. Using SQLDeveloper to monitor long running sql and session lock blocks and Dell optimizer to optimize the query and query rewite.• Upgrading ORACLE database from 11g to 12c(12.1.0.2): using JDBC ojdbc7 driver to replace old ojdbc14.

• EM UI development: Using ORACLE Application Development Framework( ADF) to create Recovery Appliance (RA) and database backup UI interface at ORACLE Enterprise Manager (EM). Did all UI related works from UI interface design, UI component choice, taskflow design, to data control model coding. Java JSPX, JSFF, and ORACLE SQL were used most.• Backend development: Oracle SQL query tuning for Recovery Appliance UI display.• Regression test automation: Wrote automatic UI test cases within Selenium framework.

• Alert service: Using IBM RAD developer to create a web service to accept alert from XTRAC; passing incoming message to MQ; EJB MQ handling incoming message from the MQ; filtering alert message and sending it to FENS web service by using SOAP.• Centera document purge: Using newest Java multi thread ExecutorService to invoke fixed number of threads based on back end system capacity to quickly handle huge EMC Centera document purge need.

• Database ETL: Created multi-threaded JAVA framework to handle different customer meter types, sum different records to do batch insertion. This architecture achieved over 45000database records insertion per second. Used Spring AOP to trace method call details.• Database data analysis: Wrote data integrity report by demand. By analyzing meter data using statistics functions such as deviation, and cube, detected potential meter’s data problem.• System test automation: Wrote a complex stored procedure to simulate real time data feed and to detect any possible system error. This was used as automatic system regression test. • SPRING, HIBERNATE framework: By creating a sample database query web GUI, compare J2EE and .NET web performance and user experience. • Data report engine: Created a multi-threaded JAVA framework to handle various type data report needs. A new type of report can be easily plugged in the framework. Various reports can be sent to various recipients based on configuration. The back end SQL used SQL server Analytic functions.• Database schema design: Abstracted complex power industry customer and its meter to two categories: entity (customer, facility, building, etc) and resources (meter, meter group, etc) and their association.• New customer web based GUI: Used Sqlbulkcopy to load Excel workbook to SQL server database; Used ASP.NET 3.5 with C# technology such as dropdown, checkbox, radio button with AJAX to develop use friendly interface to edit uploaded Excel workbook. • Baseline Engine: Backend C# application server: parse baseline rules, query meter interval reading, calculate baseline based on parsed baseline rules, and insert calculated results to database table using Sqlbulkcopy.

• Moving file smartly: The log file balance sprayer was developed to make 6 log processors evenly handle all logs. The system checked the number and last modified time of file in incoming directories and destination directories. Based on predefined sprayer rules, the files were moved to specific destination directories. The java 1.5 was used.• Oracle report query performance tuning: To speed up the data aggregation (group by) for 300 million entries, many measures such as sub partitions were used to aggregate the query and union results. Materialized view with fast update and query rewrite were also used to speed up the query. Oracle parameter such as large_pool_size tuning for better performance.• Oracle transaction speed tuning: Tuned Oracle insert/update performance for a table with type “LONG RAW” and “BLOB”. After changing column type “LONG RAW＂to “BLOB” and used 4-way hash partitions, the application performance gained 10 times faster (Long Raw type cannot be partitioned).

• JAVA server side framework: Built a general used framework which can be used by multiple projects for Margin Risk group. JAVA 1.5 thread pool is used to allocate fixed number of threads defined in configuration file; JMS (both sync and async) is used to connect MQ to do independent work at each thread. The configuration file used XML format and was parsed using XML DOM parser; each project can extend this general parser with specific application configuration. A socket is created to control the frame work server: user can send “work”, “idle”, and “stop” from any host to control the system. A timer thread is used to start/idle the work threads at predefined time and intervals. A general parser class was created to handle mainframe COBOL generated fixed length format. Log4J was used to log INFO, DEBUG, ERROR, and FATAL log. A general Interface is used in framework to connect application project easily.• Order/Execution gateway server: Converting current trade order/execution gateway (From Mainframe VSAM file to Solaris Oracle table) from C++ code to JAVA code: Directly using the above framework to implement business logic: sending time stamp of current order in database to Mainframe via MQ, main frame queue manager trigger an application which translates vsam files to a message and reply to Linux side queue manager. Java code used fixed format parser to parse message, and the JAVA JDBC batch and Oracle PL/SQL procedure to insert batch data to Oracle table. ORACLE 9.2 thin driver was used and a connection retry function was developed to handle connection failure; ORACLE connection pool is used and each thread has a logical connection.

• SAM-QFS management GUI design: Designed multiple tier product system: allows customers to access web-based Java GUI server from anywhere, GUI server host to connect SAM-QFS API host via JNI and C RPC server, and SAM-QFS API to connect SAM-QFS core via shared memory, memory mapping. • SAM-QFS management web-based Java GUI: Developed shared SAM-QFS File System GUI using LockHart, JATO, JAVA script, Servlet and JSP, and XML. JATO (J2EE Assisted Takeoff) and Lockhart technology developed HTML form tag library and used MVC pattern to implement J2EE Servlet (basically JSF internal use). • SAN development: Set up SNMP MIB to monitor SAN health state using SNMP GET and TRAP. • SAN topology object model: using togetherJ and object-relational mapping using JDO (Java Data Object).

• Click stream process: Wrote the code to process web user log: to parse for user session, to brand for category, and to profile for prime product, category, and channel using C++. Wrote PERL script to extract, clean, filter, sort, join, and transform data using ORACLE ETL tool TORRENT. • Data warehouse: Designed a web user profile data warehouse star schema. Tuned Sql performance, created and managed table space partition, RAID5, disk volume, physical data file, wrote PL/SQL scripts to do complex query optimization, performed cubic analysis. • ORACLE (8.0.5) for Vignette Story Server: Designed a database schema using SQL designer/2000. Wrote ETL C++ code to load data from other database and text files to Vignette Story Server database.

Online credit card verification using C++.