International Journal of Scientific Research and Engineering Development

The exponential growth of digital communication has positioned email as the primary medium for professional correspondence, simultaneously imposing severe cognitive load and information overload on modern workers. This paper presents Mavinmail, an advanced cross-platform email assistant engineered to automate email processing and contextual summarization. Architected as a cloud-native monorepo utilizing Turborepo, Mavinmail bridges a Reactbased browser extension and a Next.js dashboard through a unified Node.js/Express backend. To achieve latencyresistant, high-accuracy context generation, the system employs a Retrieval-Augmented Generation (RAG) pipeline powered by Pinecone Serverless vector indexes and Cohere embeddings, routing generation tasks to Large Language Models (LLMs) such as Google Gemini via the OpenRouter API framework. Asynchronous background processing is strictly governed by Redis and BullMQ, preventing main-thread blocking during concurrent email ingestion. Experimental evaluation demonstrates that Mavinmail’s decoupled architecture delivers a client-facing acknowledgment in under 150 ms while the full AI generation pipeline completes asynchronously in the background, providing a robust and scalable solution to digital communication overload.

Ronak Sanjay Rahane, Vyankatesh Ratnakar Kamod, Prathmesh Ravindra Patil, Gargee Shekhar Buva, Vrushali S Nikam,"Mavinmail: An Intelligent Email Management System Using Large Language Models and Retrieval-Augmented Generation" International Journal of Scientific Research and Engineering Development, V9(3): Page(207-217) May-June 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.