Projects

StyleVec API
Clothing search API with vector embeddings, similarity search, segmentation, and MCP integration for AI assistants.
StyleVec API
SavorSync
Cultural cooking platform with Langchain AI workflows, DSPy optimization, and user-driven UI/UX improvements.
SavorSync
Leo Explore
Autonomous frontier-exploration stack (SLAM, K-means frontiering, adapted A*, MPPI) validated on TurtleBot3.
Leo Explore
Calorie-Counter
Simple interface to track daily foods and calories with a user-supplied scale.
Calorie-Counter
Real-time 2D Object Recognition
Region-based object recognition in video using moments, percent fill, and KNN/L2 metrics with a small training database.
Real-time 2D Object Recognition
Content-Based Image Retrieval
Feature-based image matching using color, texture, regions, and deep features; compares images with metrics like histogram intersection.
Content-Based Image Retrieval
Mapping Boston Streets using an Autonomous Car
Urban LiDAR mapping with LEGO-LOAM on NUANCE vehicle; optimized pipeline for our dataset.
Mapping Boston Streets using an Autonomous Car
Optimizing Autonomous Vehicle Positioning Using IMU and GPS
Sensor fusion analysis of IMU and GPS to estimate pose and velocity on campus runs.
Optimizing Autonomous Vehicle Positioning Using IMU and GPS
Intelligent MPPT Implementation
Comparison of P&O and ANN-based MPPT on a boost converter for improved tracking and stability.
Intelligent MPPT Implementation
Road Surface Segmentation
Training strategy to improve cross-dataset generalization for road surface segmentation using deep learning.
Road Surface Segmentation
Customized Region Proposal based Traffic Sign Detection
Lightweight traffic sign detection using a custom region proposal method that reduces proposals vs. selective search.
Customized Region Proposal based Traffic Sign Detection