Computational Science & Engineering
Urban Autonomous Taxi Simulation
- Agent-based SimPy simulation with NYC Taxi and OpenStreetMap data
- GCNN-DDGF demand forecasting with 49.6 RMSE
- 5% increase in trip fulfillment and 38% fewer failures
Pneumonia Classification Project
- Hybrid Bayesian-CNN with ResNet50 and EfficientNet B0, F1 up to 73.3%
- LoRA adaptation reducing trainable parameters by over 90%
- Benchmarked fine-tuning, transfer learning, and uncertainty-aware modeling
NASA JPL MOID Regression Analysis
- Stepwise and Lasso regression in R, adjusted R² of 0.366
- End-to-end preprocessing, diagnostics, and model validation on astronomical data
Quantitative Finance
Thesis and Capstone: LLMs for Market Forecasting
- BERT-based sentiment modeling of FDA announcements
- Temporal Fusion Transformer and graph models for return prediction
- Weak supervision for knowledge graph construction and event modeling
Crypto Statistical Arbitrage
- PCA eigenportfolios and autoregressive modeling across 120+ cryptocurrencies
- Sharpe ratio of 0.54 and maximum drawdown of -3.19%
- Dynamic signal mapping, backtesting, and risk-controlled portfolio simulation
Integrated AI-Based Discovery and Innovation
- LSTM and Deep Q-Learning ensemble portfolio optimization
- Outperformed equal-weighted portfolios by 200%, SPX by 100%, and QQQ by 50%
- Reinforcement learning for institutional portfolio rebalancing at scale
Industry & Other Projects
Amazon Robotics Senior Design
- Gradient Boosting, SARIMAX, and empirical demand models for safety stock optimization
- 4.67% reduction in service reports and 4.05% reduction in costs
- Integrated predictions into a Tableau interface for decision support