Oct 2021 – Feb 2023 · 1 yr 4 mos · New York, United States

• Designed and implemented a real-time data pipeline to maintain 500M+ rows document processing data by integrating data from 15 primary data source.
• Generated automated task model in snowflake for client invoicing, saving 16 hours of manual work each bi-week.
• Created client snapshot checker dashboard in tableau to alert mismatch data between internal and external, reduced 95% wrong messages sent to clients.
• Ingested streaming data across multiple google sheets using AWS Glue and python and increase in performance.

Feb 2020 – Oct 2021 · 1 yr 8 mos · Rockville, Maryland, United States

• Developed the ETL data pipeline to manipulate 20GB+ new treasury order data, saved 20% of time cost
• Identified the manipulative behavior in the treasury security market and optimized the algorithm with dynamic thresholds, reduced 60% false-positive cases
• Depicted the instances of abnormal large orders with anomaly detection algorithms and statistical analysis including Three-Sigma Limitations, increased the profits for 12%
• Analyzed the new treasury order data feed in 4+ different dimensions including ventures type, amount, and time etc.
• Visualized the trading volume trends in top 20 price level bid/ask quotes in Tableau, presented it to the stakeholders and it was well-received by them
• Managed the software development life cycle of multiple patterns and ad-hoc requests with quality control teams

Aug 2019 – May 2020 · 9 mos · Washington D.C. Metro Area

• Developed a shiny application for automating and simplifying data segmentation and fulfillment processes for CRM Platform and collected data from multiple electronic health record platforms by building data pipelines in R.
• Delivered reports and Tableau dashboards for clients and provided insight about the health condition of black men in the DC area to improve the healthcare assistance process.

Jun 2018 – Aug 2018 · 2 mos · Beijing City, China

• Collaborated with a project manager and engineers to build a platform for a client, one of the biggest petroleum company in China, organized their history record about personnel placement using machine learning models
• Applied K-means as dimensionality reduction and built a Random Forest model with 85% accuracy to classify above 100M files through the structure data (title, date, pages and department) connecting with SQL Server
• Independently extracted named entities (employee’s name, position, company) in Python by using advanced NLP method and regular expression, the platform accelerated to 50k files per minute

Jun 2014 – Jul 2014 · 1 mo · Xi'an, Shaanxi,China

• Tuned a linear regression model to evaluate credit risk for small business companies and entrepreneurs, and the model correctly picked out over 90% clients who are with low credit, which increased 5% accuracy
• Prepared formal credit packages for approval presentation, tracked the process of credit evaluation reports and maintained effective communication with multidepartment