I engineer production-ready ML systems that solve complex, real-world bottlenecks. I specialize in deploying multi-model architectures, building robust RAG-powered pipelines, and engineering custom anomaly detection layers for edge and cloud environments.

Languages & Databases

ML, LLMs & Agentic AI

Web, Cloud & DevOps

TensorFlow LangChain FAISS Groq Llama 3.3 Docker GKE

An anomaly detection system built for critical infrastructure security.

• Engineered a custom rolling variance layer (3,300 parameters) to catch stealthy replay attacks that successfully evade standard LSTM Autoencoders.
• Architected a RAG pipeline to ground operator guidance in ICS security literature, configuring RAGAS faithfulness scores below 0.60 to trigger automated security escalations.
• Deployed the multi-user Streamlit web application to the cloud utilizing Docker and Google Kubernetes Engine (GKE).

LSTM CNN Random Forest FastAPI React

An end-to-end agricultural decision system designed and shipped within 24 hours.

• Coordinated three heterogeneous models (yield prediction, disease detection, crop recommendation) into a single, unified FastAPI service.
• Containerized with Docker and benchmarked against real-world agricultural and meteorological datasets.

🏆 Hack-Olympic 2026 · 2nd Runner-Up (National Field) · ₹20,000 Prize

Flask DeepFace OpenCV

An automated attendance tracker optimized for resource-constrained environments. Performs frame-level liveness detection to distinguish live users from photograph and video replay attacks. Runs entirely on local hardware for low-latency CPU inference.