Edwin Ma

Trained a model on 300,000 unique faces over 7 datasets to accurately recognize and encode faces using the Viola-Jones Object Detection Framework and adaptive boosting Displayed model improvement statistics Confusion Matrix, F1 Score, and ROC enhancing model performance and accuracy. Incorporated AI-based emotion recognition, incorporating CNN techniques and transfer learning to interpret facial expressions.

WORK IN PROGRESS!

Placed in the top 8% of 9,200 competitors in a trading competition, developing a trading bot using linear regression, mean reversion, and moving averages to trade in-game currency (seashells) for profit. I Utilized JSON for iterative data storage and CSV files to analyze historical data for predictive modeling.

Awarded "Most Innovative Product" and "Socially Conscious Product," in the OC Maker Invention Competition. Worked with a team of 4 designing a carbon absorber to reduce pollution targeting the factories’ smokestacks. Presented ideas with enthusiam, logic, and a 3D model.

Automatically solved Sudoku boards of various sizes rapidly using AI Heuristics, Constraint Proprogation, and Backtracking. Top 10% of class.

Developed a 2D platformer game, proposing and assigning tasks through QA testing, Git, and Discord. Programmed game mechanical and features and playtested incrementally in Unity with C#

Designed an architecture and framework for the full-stack structure for a team of 4. Implemented a data store, retrieval, and micro web system through Flask.

92nd Percentile in National Cyber League Competition. Topics include network forensics, enumeration, exploitation, cryptography, and code tracing.

Received Honorable Mention from the National MIT App Inventor Program for creating an app connecting entrepreneurs with resources and workers.

Designed a tetris bejeweled hybrid game in the python shell and converted it into a Graphical User Interface (GUI).