I build ML systems that solve real-world problems, from securing critical infrastructure to helping farmers make smarter decisions.

Languages

ML & Data Science

Web & Tools

LSTM CNN Random Forest FastAPI React

End-to-end agricultural decision system built during a national hackathon. Combines three specialized models — crop yield prediction (LSTM), disease detection from imagery (CNN), and crop recommendation (Random Forest) — into a unified real-time inference pipeline. Farmers get actionable insights from minimal input.

Hack-Olympic 2026 · 2nd Runner-Up · ₹20,000

TensorFlow LSTM Autoencoder Streamlit

Anomaly detection system for critical infrastructure security. Introduced a rolling variance layer to catch stealthy replay attacks that standard models miss. Processes 174-dimensional telemetry data with an interactive Streamlit dashboard for real-time visualization.

Flask DeepFace OpenCV

Automated attendance tracking via facial recognition with built-in liveness detection — no spoofing, no proxies. Clean, real-time pipeline from camera feed to attendance record.