Projects

End-to-end perception-to-control framework that uses 3D Gaussian Splatting and differentiable physics to identify object properties and plan contact-rich robotic manipulation.

Temporal Fusion Transformer–based system for forecasting S&P 500 returns from mixed-frequency macroeconomic and market data, with ARIMAX and LSTM baselines.

Full-stack autonomous quadrotor system combining computer vision, SLAM, optimal control, reinforcement learning, and waypoint navigation.

Real-time system that tracks a ping pong ball with YOLOv8 on a Raspberry Pi and steers a servo-driven goalie by predicting 3D trajectories.

An IIoT-enabled smart factory cart with end-to-end stack from edge control on a Raspberry Pi to central fleet analytics dashboards.

Transformer-based descriptive question answering system using a fine-tuned BERT model and sentence-ranking pipeline, published at an IEEE conference.

Comparative study of DCGAN, Progressive GAN, and diffusion models on faces, synthetic colored squares, and a custom animal image dataset.

Selfie segmentation and realistic lens-blur simulation using pretrained segmentation and monocular depth-estimation models.

High-resolution image representation using sinusoidal representation networks and a custom activation variant.

Sequence model that generates symbolic music with controllable temperature and musical parameters.

Systematic study of VGG-based perceptual losses under shifts, noise, and depth-aware atmospheric turbulence, with a new turbulence-robust weighting scheme.

Fine-tuning ViT-B/32 on a class-balanced photovoltaic infrared defect dataset to achieve high-accuracy multi-class classification.

Custom ResNet-36 architecture and SmoothReLU activation that improve ImageNet-scale training from scratch with minimal parameter overhead.