Metadata Engineering

Ads Backend(Yahoo! DSP Advertising)

• Worked on component tasks of DSP Budget Control Server, optimization tasks of DSP Bid Server and integration tasks with Yahoo! Gemini Native Ads

• Implemented QuickStop Throttler which successfully reduces the load of the quick-stop events by ~97% sent from DSP(Demand Side Platform) Budget Control Server to DSP Optimization Server

• Implemented QuickStart and QuickStop functionality which successfully handles daily budget allocation for Yahoo! Gemini Native Ads

• Optimized DSP budget control system database queries and successfully improved database query performance by reducing 1/3 of the time taken by original database query

• Designed and implemented the project of reducing DSP Bid Server CPU utilization by at least ~10% (as committed from avg ~63% to ~47%) across the COLOs by optimizing data encryption code, which also helped reduce the p99 bid latency by ~10%

• Designed and implemented dynamic calculation of the time budget for the I/O calls (e.g., time taken of calls from DSP Optimization Server) to have the better control from the E2E latency point of view

• Designed and implemented Control Server monitoring system which supports monitoring data metrics from partition and cluster level

• Participated in the design and implementation of building up docker stage testing environment for DSP Budget Control System

• Designed and implemented the refactorization of Logging System

Product Analytics

• Backend Server(Kafka, ElasticSearch, Logstash, GraphQL, Presto, Druid, Docker, MyBatis, Kotlin)

• Led the design and implementation of the search module built on top of ElasticSearch and Kafka

• Built the file upload feature using Kotlin, Spring Boot, GraphQL and S3

• Implemented Auth0 authentication for the SaaS platform with NodeJS and ReactJS

• Integrated with Jupyter Notebook through Jupyter Server Extension and RESTful API with Python and Kotlin

Ooma-Nest Integration

• Backend Server(Kafka, Jetty, Docker, Kibana, OAuth2.0, Node.js, Java)

• Worked on the HomeSecurity project and implemented support functionality for Bluetooth low-energy sensors

• Developed backend server using Apache Tomcat/Jetty, MySQL, Docker, Kibana and Kafka

• Integrated Ooma Telo with Nest Smoke Detector using NodeJS

• Analyze datasets from government and non-government websites, and model and track the corporate structure of businesses via government filings and online research.

• Make inferences about the financial health of companies based on web metrics and verifiable government filings and news articles, and analyze the extracted data, train the algorithms, and then apply the algorithms to thousands of companies.