Projects
Attention Kernel Portability Across GPU Architectures
A benchmark of forward-prefill and single-token KV-cache attention across six NVIDIA GPUs, asking whether the fastest backend on one device is still the fastest on another. A winner counts only when it clears both a bootstrap confidence interval and a 10% margin; under that bar the backend changed in 27 of 64 separated prefill comparisons across the Ampere, Ada and Hopper parts.
DRISHTI: On-Device Assistive Navigation for the Visually Impaired
A wearable assistive system pairing a Raspberry Pi belt with an Android app that runs Gemma-3n vision-language inference, ArcFace face recognition, and sensor-fusion fall detection on-device, behind a fully voice-driven interface.
CXRGenie: Multimodal Chest X-ray Diagnostic Assistant
A modular agentic AI for chest X-ray interpretation built on GPT-4o-mini, integrating segmentation, classification, grounding, VQA, and report generation with a RAG-backed vector DB for retrieval-augmented, human-in-the-loop reasoning.
2D-to-3D Coronary Artery Reconstruction & Stenosis Detection
Algorithms for reconstructing 3D coronary geometry from X-ray angiograms, improving stenosis-detection accuracy and cutting computation time via pseudo-transient vFFR.
CVS Classification & Hepatocystic Anatomy Segmentation
Optimized deep models for laparoscopic cholecystectomy: video-based Critical View of Safety classification and anatomy segmentation to assist surgical workflows.
3D Super-Resolution: GANs vs. Diffusion vs. CuNeRF
Comparative study of z-axis 3D MRI super-resolution across GAN, diffusion, and implicit-field methods, analyzing the quality–latency trade-offs for clinical deployment.
FusionDecompNet: Multimodal Alzheimer's Diagnosis
A multimodal multitask framework for Alzheimer's classification (NC/MCI/AD) via joint MRI–PET fusion with auxiliary modality reconstruction.
Defect-Aware Denoising & Deblurring Network
An encoder–decoder with spatial-channel attention and non-local blocks for defect-preserving image restoration.
Metal Additive Manufacturing Defect Detection
A spatial-attention segmentation model to detect spatter-particle defects in X-ray CT, improving defect detection for metal additive-manufacturing quality control.
sEMG Modeling & MUAP Decomposition
Collected and analyzed single-channel sEMG during hand-reaching tasks; applied signal processing and ICA to extract motor-unit recruitment patterns during muscle contraction.