Selected Projects
End-to-end projects: research and deployed prototypes
Telecom KPI Anomaly Detection Using AI Agents
Multi-agent AI system with LangGraph, MCP, and anomaly detection for real-time KPI monitoring across 100+ sites.
Image-to-Video Generation of Anime/Comics
3D latent video diffusion model with 71.8% SSIM improvement over baseline interpolation.
Retrieval Augmented Generation from YouTube for Long-Form QA
RAG pipeline transforming video data into searchable knowledge bases with LangChain and LLMs.
Meal Nutrition Analysis via Multi-Modal Deep Learning
CNNs, LSTMs, and demographic embeddings—34% accuracy boost over benchmarks.
Autonomous Navigation with RRT* and A* Path Planning in ROS 2
Motion planning pipeline with A*, RRT*, and pure-pursuit control.
LiDAR-Based Human Detection and Multi-Target Tracking
2D LiDAR perception pipeline with clustering and multi-object tracking in ROS 2.
Robust RL for Mixed Autonomy Traffic Systems
PPO and TRPO agents with 20% throughput increase and zero safety violations.
Face Tracking Robot
4-wheeled mobile base with computer vision for real-time face detection and person following.
Brain Tumor Segmentation
W-Net and deep neural networks for medical image segmentation.
Vision Transfer Learning
DeiT-based image classification with augmentation and fine-tuning.
Cartpole using Q-learning
Reinforcement learning implementation for cartpole control.
Health Monitoring & Heart Stroke Prediction
IoT-based health monitoring system with ML prediction.