• Instagram Ads Team
• News Feed Growth Team
• Collaborated in building a mobile DSP infrastructure scaled for 1000 ad campaigns on both frontend (AngularJS) and backend (Go).
• Launched an Android, iOS, Unity, and JavaScript SDK for mobile event tracking using RESTful API (Moloco VAN).
• Built an internal chat bot for campaign management with ad-hoc features such as summarizing KPIs from cloud database (BigQuery).
• Automated and simplified over 35% of the campaign operation work flow by scripting tools such as price model generator.
• Developed various machine learning based prediction models using raw data scraped from the MLB Gameday website using R and Python.
• Designed unique features, automated the model training process, and simplified cross-validation and ROC analysis for evaluations.
• Created an API for predicting the next pitch ball type using pre-trained models for MLB team managers.
• Designed a baseline prediction model for legal analytics in judicial decisions made in U.S. Court of Appeals for the Federal Circuit.
• Built data ETL pipelines for constructing a relational database from world-wide-web, legal documents, and existing databases.
• Intro to Machine Learning (Jan 2017 ~ May 2017)
• Data Structures (Sept 2016 ~ Dec 2016)
• Oral Presentations (Jan 2016 ~ Dec 2016)
• Wrote an optimization program in JavaScript for visualization of a network of multi-dimensional-nodes on HTML Canvas.
• Devised functions for producing optimized coordinates for each node and edge curve point by calculating collisions and noise levels.
• Served as a military demolitionist in the Engineering Battalion of the 21st infantry division for 2 years.
• Awarded the Best Squad Leader award from division commander during 2015 spring training camp.
• Assorted product sales data (~$3.5billion/yr) from the database, conducting linear regression analysis using Excel for sales projections.
• Summarized recent sales trends, estimated future revenues, and identified potential target clients (sales flux over 15%) using pivot tables, creating a business report along with graphical models to represent the performance of company products and sales teams on weekly basis.
Computer Science, Applied Mathematic & Statistic Major
Social Dating Matching Prediction
Modeled classification predictor using Speed Dating Experiment of 21 waves to classify dating candidates as potential or non-potential match based on demographic attributes and partner preferences using three different classification algorithms (Perceptron, SVM, KNN).