Projects
A selection of my work across ML, data systems, and distributed systems
Predictive Modeling — Roosevelt National Forest
Built a 54-dimensional classification model on 580k records achieving 95.6% accuracy using Random Forest. Applied SMOTE for class imbalance and PyTorch MLP for non-linear interactions.
MicroTorch — Custom Neural Network Framework
Built a neural network framework from scratch in Python/NumPy with automatic differentiation and backpropagation. Achieved >95% accuracy on MNIST dataset.
Full-Stack Database Application
Designed and implemented a complete database application with EERD modeling, relational schema design, and a full-stack web interface with SQL backend. Optimized queries for performance.
Distributed In-Memory Cache
Implemented a distributed cache system with consistent hashing and LRU eviction, achieving 99th percentile latency under 10ms for 100K ops/sec.
Production dbt Data Pipeline
Designed and deployed production dbt models handling critical data transformations at Delta Air Lines. Built automated testing, documentation, and CI/CD integration.
ORB-SLAM Computer Vision for 3D Mapping
Implemented ORB-SLAM for autonomous drone 3D mapping with sub-centimeter localization accuracy. Integrated QR code automation for warehouse inventory tracking.