Garvit Agarwal

Designed and deployed production-grade Retrieval-Augmented Generation (RAG) systems using Python, FastAPI, and vector embeddings, reducing response latency by 40% and improving answer relevance by 30%. Built scalable RESTful APIs for AI-driven platforms (Market Research Intelligence and enterprise Trial Chatbot) and designed a secure end-to-end document intelligence pipeline with summarization, intelligent chunking and embedding generation reducing manual effort by 60%.

Designed and optimized distributed ETL pipelines using PySpark, Databricks, and SQL to process multi-source datasets (10M+ records), transforming raw data into analytics-ready data marts; awarded Best Delivery Excellence Award. Improved reporting accuracy by 60% and reduced pipeline runtime by 35% through performance tuning, query optimization, indexing strategies, and automated validation checks in SLA-driven production environments. Automated cross-platform data validation between Snowflake and Power BI using Python-driven SQL workflows and Power Automate, reducing manual QA effort by 80% and accelerating delivery timelines by 25%.

Developed and optimized CNN-based deep learning models for maritime ship classification in a defense surveillance system, achieving 92.97% classification accuracy across multiple ship categories. Executed the complete machine learning lifecycle including data preprocessing, augmentation, model training, hyperparameter tuning, and evaluation on 5,000+ labeled images, leading to selection among the Top 75 projects at Swavlamban 2023.