Manav Kaushal

Architected an AI-powered educational assistant using RAG (Retrieval-Augmented Generation). Implemented custom prompt chains to deconstruct complex codebases into digestible learning modules.

Designed and deployed a generative AI prototype integrating Gemini APIs with RAG (Cohere embeddings) to enable scalable document analysis using FastAPI, LangChain, and React.

Developed an automated ETL pipeline for state budget analysis, performing data cleaning, schema transformation, and reporting to improve accessibility and policy insights.

An intelligent tool for generating tailored cover letters and resumes. Optimizes content for ATS systems using advanced language models to increase interview chances.

AI-powered disease outbreak early warning system that fuses health clinic reports, pharmacy sales spikes, social media trends, and climate data. Uses time-series anomaly detection to predict outbreak probability before traditional surveillance catches up.

Computer vision and NLP surveillance intelligence platform for police operations. Combines YOLOv8n object detection (31ms inference), crowd density analysis, anomaly detection, and a RAG pipeline with FAISS vector store — all served via a FastAPI backend and a React command center dashboard.

Developed a predictive maintenance system for heavy-duty vehicle air pressure systems. Achieved 98% recall using XGBoost and SMOTE for handling extreme class imbalance.