Projects

• Computer Vision Satellite Image Analysis

  Analysis of cargo movement in high forestation areas using computer vision. General aspects: depth estimation, multi-temporal object tracking, precise semantic and instance segmentation of ground features, and foreground change detection and multi-class object classification, specifically designed to identify diverse cargo types, vehicles, and associated activities within densely vegetated or complex terrain. Intended purpose is to provide actionable geospatial insights that can identify and monitor human trafficking operations in high-risk areas.

• AI Therapist

  Using natural language processing to make therapy more accessible, affordable, and convenient. Technical aspects: made a Seq2Seq model for therapist-patient interactions, tokenization for keyword identification (greetings, emotions), data cleaning and sentence-level pairing of questions (x) and answers (y), padding/trimming for uniform input length, using two LSTM networks within the Seq2Seq architecture, incorporating a transformer model with Multi-Headed Attention to grasp full conversational context and address ambiguity, and mapping words by similarity for enhanced understanding.

• Monte Carlo Poker Estimation

  Using Monte Carlo simulations to estimate the probabilities of different boards in poker. Player has options to randomize their hand or enter a given hand at each stage of the game as well as enter the number of players still at the table. Project collaboration with my brother,
  Cappillen Lee

• Spotify Blender

  Given two users, this app uses the Spotify API to list all songs from users' profile playlists then has an algorithm find similar tracks and same artists to create one playlist. Ccass algorithm is used to calculate musical compatibility between two users based on genre, shared artists and tracks. Project collaboration with my brother,
  Cappillen Lee