Full-Stack Developer · MERN Stack · Building at the intersection of Web & ML

I'm a 2nd-year B.Tech student who builds full-stack products — not just tutorials or clones. I specialize in MERN stack development, and I'm currently expanding into Machine Learning, applying it to real-world problems like attendance automation and resume parsing.

I care about clean architecture, real user experience, and shipping things that actually work.

• 🚀 Just shipped: OnRecord — AI attendance system using face + voice biometrics · Live →
• 💪 Strong in: React, Node.js, Express, MongoDB, WebRTC, WebSockets
• 🧠 Exploring: Python ML pipelines, speaker embeddings, computer vision
• 🤝 Open to: Internships, collaborations, and interesting problems

Replaces manual roll-calls with a biometric verification system. Instead of asking "who's present?", it answers "who can be verified?"

• Detects multiple faces per image using dlib — generates 128-d face embeddings matched via trained SVM classifier
• Real-time speaker verification using resemblyzer d-vector embeddings + librosa acoustic feature extraction
• One-time biometric enrollment (face + voice) stored securely via Supabase auth + database
• QR code-based instant course joining — no manual data entry for students
• Confidence scores logged per verification with exportable CSV attendance reports

Stack: Python Streamlit dlib scikit-learn resemblyzer librosa Supabase Pillow Pandas

View Repository →  ·  Live Demo →

A fintech web app for portfolio tracking, market watchlists, stock analytics, and simulated order management.

• Built real-time dashboards and interactive charts with React
• Designed scalable backend APIs with MongoDB for user, portfolio, and transaction data
• Implemented authentication flows and modular trading interfaces
• UI/UX inspired by modern brokerage platforms

Stack: React Node.js Express MongoDB Chart.js

View Repository →

A full-stack communication platform with video conferencing, P2P calling, and instant messaging.

• Implemented low-latency video/audio using WebRTC peer-to-peer architecture
• Built real-time signaling and chat using WebSocket technology
• Developed scalable backend with MongoDB for session and chat persistence
• Responsive meeting interfaces with modular frontend components

Stack: React Node.js WebRTC WebSockets MongoDB

View Repository →

Parses resumes and scores them against job descriptions using NLP — helping candidates identify gaps before applying.

Stack: Python Streamlit NLP spaCy

Frontend

Backend

Database

ML / Data