_{B.Tech · Galgotias College of Engineering and Technology}

I'm an engineering student working across AI/ML, computer vision, backend systems, and embedded technology. My focus is on taking real problems — not toy datasets — and building systems that actually hold up: explainable models, measured latency, and interfaces people can use.

My core technical anchor is a semiconductor wafer defect detection system, currently the subject of a patent application, which has shaped the direction of most of my subsequent project work and hackathon pitches.

Deep learning system for semiconductor wafer defect classification, built with a focus on interpretability and deployability rather than raw accuracy alone.

• ResNet18 classification pipeline with focal loss, trained on the WM-811K dataset across eight defect classes
• Grad-CAM based explainability layered with geometric boundary validation and rule-based inference for more trustworthy predictions
• ~12ms CPU inference latency, making it viable for real-time / edge deployment
• Confidence scoring and an interactive prediction dashboard on a full-stack web interface

A 12-module AI-powered study platform designed to make self-directed learning interactive rather than passive — spanning an AI Teacher, PDF engine, quiz engine, flashcards, voice AI, and a placement-prep module.

• Built on React/Vite (frontend) and FastAPI (backend), with the Anthropic API and ChromaDB for retrieval
• OCR-powered question extraction and an image-to-explanation workflow
• Designed for offline-first, local AI assistance

An interactive personal portfolio showcasing my AI/ML, computer vision, and embedded systems work.

• Built with Vite + React, React Three Fiber, Framer Motion, and Tailwind CSS v4
• View it live →

I treat DSA as a standing practice, not a phase — breaking problems down, comparing approaches, and optimizing for both correctness and efficiency.

Open to conversations on AI, computer vision, systems design, or interesting problems worth building for.

Portfolio · LinkedIn · Email · GitHub

_{Building with curiosity. Learning with consistency.}