Projects
NVIDIA Robotics for Developers
22 hr course on NVIDIA Robotics
​
https://nvdam.widen.net/s/brxsxxtskb/dli-learning-journey-2009000-r5-web
NVIDIA Isaac Groot
NVIDIA Deep Learning Institute: Intro to Robotic Simulations in Isaac Sim
UC Berkeley ML&AI Project: Comparing Classifiers
In this project, I looked at sales data to predict the probability of a sale to specific customers. Dummy regressor, logistic regression, k-nearest neighbors, decision trees, SVC with linear, SVC with poly, SVC with rbf, and SVC with sigmoid were used to classify a dataset with 16 features.
​
After hypertuning, a 90% accuracy score was achieved with k-nearest neighbors being the best all-around model for this dataset.
​
Reinforcement Learning by David Silver
Introduction to Reinforcement Learning by David Silver
​
https://www.youtube.com/playlist?list=PLzuuYNsE1EZAXYR4FJ75jcJseBmo4KQ9-
UC Berkeley ML& AI Capstone: Fruits Classification
Fruits Classification using 10,000 images of bananas, strawberries, apples, grapes, and mangoes. Conv2D, MoblieNetV2, EfficientNet, ResNet152V, Inception V3, Xception, and ConvNeXtBase pre-trained models were used and results analyzed.
​
After hypertuning, a 95% validation accuracy score was achieved using MobileNetV2.
UC Berkeley ML&AI Project: Used Car Sales
In this project, I used a Kaggle dataset with used-car data to predict what dealerships should charge for a used car. 17 features (e.g. region, model, year, odometer, etc) were used to predict price.
​
The data was very messy with a lot of mistakes, so a lot of time was taken to do pre-processing. Even after hypertuning, a R2 score of only 36% was achieved. Deeper modeling is possible but was out-of-scope for the assignment.
​
