Portfolio
vcdash
all these vc investments and not one for me?
vcdash is a dashboard, mindmap tool, and LLM tool all fused into one that dynamically creates both text and charts on a flexible background. It was built for CxC 2025 and won third place for the RunQL challenge!
Check out the devpost: Devpost
Codecracker
winning not guaranteed, immunity from your friends’ disappointment also not guaranteed
Codecracker is a codenames solver that uses word embeddings to find the closest word vectors to a set of words to provide clue words for a codemaster. It’s a… little janky, but future improvements using more sophisticated approaches are planned!
Live Demo Here: Codecracker
Toxicity Classifier
blurred for your eyes
Toxicity Classifer is a classical machine learning project that uses multiple ML and DL models to classify comments into different categories of toxicity. The dataset is available online, and it was a part of Jigsaw & Google’s toxic comments classification challenge.
Toxic Comments Classification - Slides
Toxic Comments Classification Notebook (Warning: Extreme Language as Part of the Project)
Black Friday Sentiment Project (Big Data)
The Black Friday Sentiment project is a project that aimed to analyze general sentiment around black friday using tweets! I learnt a lot about Apache Spark, and working with big data overall! The tweet dataset I worked with had 1 million+ rows overall, and it was a lot of fun learning how to use Databricks clusters and Spark to parallelize operations. It used AWS Athena and QuickSight to pull from a S3 bucket and create a dashboard.
Uber Ridership Visualization
The Uber Ridership Visualization is a project I used to get familiar and learn about Tableau. It uses public Uber ridership data to visualize ridership patterns in the New York City area as I try to find patterns of how people take uber rides in different weather conditions and times of week/day!
HobbyNet
HobbyNet is a full-stack web app that was built with a group of web developers! The app aimed to connect locals based on their hobbies, and was built using React and PostgreSQL primarily.