Selected AI Projects

• Information Retrieval: Implemented an e-commerce search engine. Used LLM for text embedding and utilized a fair performance metric leveraging partial match for improved performance.

• Conversational Guided Meditation: Collected and processed audio and brain wave data (i.e., electroencephalogram, EEG), fine-tuned large language models (LLMs) for personalized guidance, and synthesized realistic voice output, containerized and deployed with Docker.

• Blockchain Analysis: Calculated, analyzed, and visualized total value locked, swap volume, and pool revenue for the Ekubo Protocol for multiple exchange trading pairs. Suggested a methodology to incentivize healthy growth of the protocol.

• Audio Feature Extraction: Predicted audio features by converting them to images and using Convolutional Neural Networks (CNN). Used data and metadata from Spotify (collected from API and scraped from their website).

• Impact of AI Adoption: Evaluated the performance of an AI assistant through training a Convolutional Neural Network (CNN) model for chest X-ray image classification. Discovered AI outperformed human experts overall in challenging edge cases.