[ projects ]

Research project developing inductive graph-based learning approaches to improve performance of multimodal (image + text) data representations for applications in Zillow Search. This was a joint effort by myself and fellow student researchers Adi Srikanth, David Roth, and Tanya Naheta under supervision of the Zillow Applied Science team and NYU.

Contains work completed by myself, Adi Srikanth, and Jin Ishizuka validating and extending experimental results from SimCSE: Simple Contrastive Learning of Sentence Embeddings by Gao, Yao, and Chen from Princeton University. Our work involved replicating experimental results from the paper, developing a neural sentiment classifier on scraped Twitter data to compare the performance of BERT and SimCSE-generated sentence embeddings as inputs, and evaluating BERT-based and SimCSE-based sentence encodings against each other using feature permutation analysis during sentiment classification.

Click [Read More] below to access source code and a full summary on data sources, methods, and results in the README of the project's GitHub repository.

Project exploring various ML-based approaches for building movie recommender system. Focus was on designing, building, and evaluating latent factor models that could train in parallel on medium-large (+1 GB) distributed datasets on a high-performance computing cluster. Models were trained on interactions data (ratings by 280,000 unique viewers on 58,000 unique movies) and evaluated based on how closely their top 100 recommendations for each user matched their true top 100-ranked movies by rating.

Click [Read More] below to access source code and a full summary on data sources, methods, and results in the README of the project's GitHub repository.

Interactive web application that uses logistic regression to sort any Spotify user’s Liked library into playlists.

Flask / Plotly-Dash web application is live on AWS and available here:
http://contextify.us-east-1.elasticbeanstalk.com/

Full-stack, interactive web application to track your goals, subtasks, and completion rates in real time. Check it out on Google Cloud Platform:
(https://busybee-264219.appspot.com/)

For best results, login with these credentials:
username: test1       password: test

Use case development and prototyping of a solution using NFL’s NextGen player tracking data to generate valuable scouting and coaching insights for NFL teams.