My past research has been focusing on application of machine learning (probabilistic machine learning, statistical, ...) and how it could be used to find hidden insights.

Past work:

1. ML approach for predicting non-matriculated international students.

I analyzed non-matriculated international students by using probabilistic Machine Learning. The tricky part about this dataset is that it would get overfitted easily since the dataset are time-series, and usually college decisions are determined in a steps-by-steps sequence (which means certain features are heavily depended on another feature). If we were to use normal classification approach, this would not work and will be overfitted.

Motivation: Since offering scholarships involves a significant cost, and resources are wasted on those who are accepted but don't end up signing the offer letter, a good way to contribute is to built an early prediction model to gain insights into this group. That way, we can leave the scholarship to the students who are actually in need.

2. Undergraduate Thesis: Deterministic features of STEM students using Machine Learning

I implemented t-SNE and PCA to understand the correlations to understand what features have the most significant impact towards the rate of leaving university in STEM students, reduce dimensions to find a good fit, and implemented a Lasso Regression on the entire dataset.

3. Serverless cloud defense architecture

I worked on a project to build a serverless IPS system in AWS using ECS, EC2, API Gateway, Sagemaker.