Hello, I'm

My Projects

• 

  Musical Genre Classification

  For the MIT class "Advances in Computer Vision" (6.8300), our group developed a state-of-the-art musical genre classifier that employs CNN + BiGRU architecture.

• 

  LyricsBERT: Musical Recommendation

  For the MIT class "Quantitative Methods for NLP" (6.8611), we developed LyricsBERT, a transformer-based model that recommends music based on lyrics.

• 

  EfficientVIT for Retail

  For the MIT class "TinyML and Efficient Deep Learning Computing" (6.5940), our group applied the state of the art EfficientVIT segmentation model to classification of retail items as a downstream task.

• 

  Fairness in Computer Vision

  Working with MIT Decentralized Information Group (DIG), I leverage Label Free Concept Bottleneck Model (LF-CBM) and LLMs to improve fairness in modern image classification models.

• 

  Detect DeepFakes

  Working with MIT Affective Computing group, I created 20+, 720p deepfake videos that were used to analyze visual and auditory perceptions by which AI models and humans use to detect deepfakes.

• 

  2023 MIT GTL x Amal Blogs

  As part of the 2023 MIT GTL Program in Israel where my team led an entrepreneurship workshop in local Amal Networks schools, we created a website encapturing our experiences in a blog format.

• 

  Stock Analyzer

  Want to see how much a stock is worth based on your key assumptions? Want to conduct this analysis using real-time market data? I developed a stock analyzing tool where one can easily calculate the fair value of a stock.

• 

  Personal Website

  The website you see right here was coded by me using Next, Express, and Node. The website is easily customizable with different functional components and custom inputs.