Attribution and Segment Team:

• Led the design and implementation of a highly available and scalable automatic report management and generation server, unifying all major company reporting requirements under a single, efficient processing system

• Architected, designed, and implemented a universal event-based alert system, integrating it with the attribution report server to streamline debugging and inter-team collaboration and ensure client SLA compliance

TV Advertising Platform Development Group:

• Designed and developed an automated attribution reporting system with Python and Spark to analyze LG TV audience reactions and conversions from ad exposures (including app launch and installation, audience engagement and acquisition metrics), delivering comprehensive reports to LG Ad Solutions USA

• Designed and built a UI dashboard and its web server with Python, Javascript, React, and Flask, streamlining attribution report request ingestion, validation, and result visualization for internal and external stakeholders

Attribution and Segment Team:

• Developed a scalable attribution reporting engine using Spark and Scala to measure the closed-loop performance of TV/OTT/Digital ad campaigns, driving web, mobile, and foot traffic

• Researched and developed strategies to assess ad effectiveness, including the impact from network, daypart, recency, and frequency, test-control group selection and lift analysis, and location visits data refinement, using cross-domain data modeling of TV viewing behavior and demographic attributes from 34 million U.S. households

• Developed data pipelines using Airflow for viewership and 3rd party web/mobile/location data ingestion and transformation, powering downstream services and improving analytics team efficiency by 20%

• CA for EECS E4764 Internet of Things - Intelligent and Connected Systems (Fall 2017)

• CA for EECS E6720 Bayesian Models for Machine Learning (Fall 2017)

• Built an auto-scaling web app backed by Flask, Elasticsearch, and AWS Elastic Beanstalk to visualize qualitative and quantitative insights of TV ads on social media, with D3 based charts, diagrams, and an embedded Google Map

• Developed a web crawler with Python and Kafka to fetch viewers’ activities from Twitter, Facebook, and Instagram

• Implemented a lexicon-based approach to perform sentiment analysis on 300K tweets across 2 months