I'm a B.Tech CSE (AI & ML) student at Hyderabad (2023–2027), building at the intersection of machine intelligence, computer vision, and modern full-stack engineering. I don't just study AI — I ship it.

From a biometric attendance engine that processes live video feeds to a deepfake detection system, my work is grounded in one principle: real systems that solve real problems.

shrish = {
    "role"       : ["AI/ML Engineer", "Computer Vision Builder", "Full-Stack Developer"],
    "stack"      : ["Python", "Django", "React", "OpenCV", "YOLOv8", "Transformers"],
    "interests"  : ["Deep Learning", "Smart Campus Systems", "Scalable Web Apps"],
    "currently"  : "Building intelligent systems & exploring LLM-powered applications",
    "open_to"    : ["Internships", "Research Collaborations", "Hackathon Teams"],
    "motto"      : "Build. Break. Learn. Repeat."
}

I approach every project as a product engineer: thinking about users, scale, and long-term maintainability — not just making it work. I believe the best engineers are those who can move fast and build things that last.

AI / ML / Computer Vision

Full-Stack Development

Data & Infrastructure

Computer Vision · Django · OpenCV · Deep Learning · Mobile Integration

A production-grade smart campus biometric engine built from scratch during a hackathon. Detects, recognizes, and verifies student faces in real time via webcam or mobile camera — automatically marking attendance with login/logout timestamps.

• Real-time face detection using deep learning models with live video stream processing
• Multi-device support — works with mobile cameras and webcams out of the box
• Admin dashboard with session management, class-wise reports, and CSV export
• Zero manual intervention — fully automated attendance lifecycle
• Built in a hackathon environment, now extended into a scalable campus product

Django OpenCV Deep Learning SQLite REST API Bootstrap

Media Forensics · CNN · Transfer Learning · Binary Classification

An AI-powered media authenticity system designed to detect AI-generated or manipulated video/image content with high accuracy. Trained on benchmark deepfake datasets using transfer learning architectures.

• CNN-based classifier fine-tuned on deepfake benchmark datasets
• High-confidence binary classification — Real vs. Fake with probability scores
• Designed with a modular pipeline for plug-in model upgrades
• Explores Transformers and frequency-domain analysis for next-gen detection

Python PyTorch Transfer Learning OpenCV CNN Flask

Full-Stack Engineering · Django · React · Payment Integration · REST API

A full-featured, scalable e-commerce application with a Django backend powering a dynamic React frontend — built to production-standard architecture with authentication, product management, cart flow, and order tracking.

• RESTful API architecture with Django REST Framework
• Secure user authentication, role-based access, and session management
• Complete cart, checkout, and order lifecycle with payment gateway hooks
• Mobile-responsive UI with optimized load performance

Django React DRF MySQL JWT Auth Bootstrap

🔭  Advancing in YOLOv8 fine-tuning for specialized detection tasks
🧬  Exploring Vision Transformers (ViT) and attention-based architectures
🕸️  Scaling Django + React with REST APIs and microservice patterns
🔐  Deepening Smart Campus Biometric Systems with anti-spoofing layers
📊  Building real-time analytics dashboards for ML model outputs
🤖  Experimenting with LLM-powered applications and RAG pipelines

• 🏆  Built a full biometric attendance system in a hackathon sprint — from idea to a working product with admin dashboard, mobile support, and CSV reporting
• 🚀  Shipped complete, functional applications under tight hackathon constraints — not prototypes, real tools
• 🤝  Experienced in collaborative, fast-paced engineering environments with cross-functional teams
• 💡  Driven by a philosophy of outcome over output — every project targets a measurable real-world impact

  

If you're building something at the intersection of AI, engineering, and real-world impact — let's talk.