Projects

Endeavors from various courses, competitions, and other personal pursuits.

* Alphabetical authorship

  1. Clustering and Predictive Modeling in Competitive Gaming (2025)
    Taro Iyadomi
    Applied unsupervised and supervised learning methods to identify behavioral archetypes in competitive gameplay data. Using k-means clustering, PCA dimensionality reduction, and classification algorithms, analyzed patterns across 250+ high-ranked players.
    Article | GitHub | PDF

  2. Reinforcement Learning for Snake (2024)
    Chengheng Li, Taro Iyadomi, Lizabeth Tukiman
    Implemented and compared tabular Q-Learning and Deep Q-Learning approaches for sequential decision-making in a game environment (Snake). Investigated the impact of state space abstraction on learning efficiency, convergence properties, and generalization performance.
    GitHub | PDF

  3. Behavioral Modeling for E-Commerce (2024)
    *Darren Hsieh, Taro Iyadomi, Ryan Kawamura, Axel Malvaez, Daniel O’Brien
    Predicted customer purchase journeys using XGBoost and deep learning, achieving 73% F1 score on real-world behavioral data from Fingerhut.
    Article | GitHub | PDF | Presentation

  4. NLP Pipeline for Legal Text Classification and Client Matching (2024)
    Taro Iyadomi
    Built an end-to-end natural language processing system for legal query classification and attorney-client matching. Fine-tuned DistilBERT transformers on legal text data and applied K-Means clustering for attorney specialization grouping. Runner-Up at ASA DataFest 2023.
    Article | Presentation

  5. Statistical Analysis of Academic Performance (2023)
    *Minhao Han, Taro Iyadomi, Marlene Lin, Meha Mukherjee, Laura Ngo, Shiqin Tan, Emre Turan
    Applied stepwise multiple linear regression to identify socioeconomic and campus climate factors affecting UCLA undergraduate GPA.
    Article | GitHub

  6. Siamese Neural Networks for Multilingual Document Similarity (2023)
    Taro Iyadomi
    Developed a deep learning architecture for semantic similarity assessment across 150,000+ multilingual educational documents. Implemented a Siamese network with shared BERT-based encoders and contrastive learning objectives, achieving 0.86 AUC on curriculum-content alignment tasks.
    Article | GitHub | Kaggle

  7. Ensemble Learning for Car Accident Severity Prediction (2022)
    *Anish Dulla, Taro Iyadomi, Nishant Jain
    Built Random Forest models with extensive feature engineering to predict car crash severity, achieving 93.5% test accuracy and top-15 placement in course competition.
    Article | GitHub | Kaggle

  8. Statistical Modeling of Opioid Use Patterns (2022)
    *Jeffrey Gutierrez, Taro Iyadomi, Jonah Jung, Andrew Schweitzer
    Analyzed drug use data across U.S. states using regression methods to model opioid epidemic spread and inform intervention strategies.
    Article | GitHub