Mrinal Setty

IEEE Research Paper (document number 9972172) – Presented in the 3rd Global Conference for Advancement in Technology (GCAT). Web based application to enable users to share data files between devices using sound, with advanced efficiency and security of peer-to-peer file sharing. Developed using C++, Python and Golang.

A full-stack Generative AI application leveraging multiple AI Agents built using the LangGraph Agentic AI Framework. The backend, powered by Flask, integrates LangChain and RAG for intelligent reasoning and contextual retrieval. The system utilizes Vector Databases, Gemini LLM and Guardrails for high-performance, privacy-safe AI automation. The ReactJS frontend delivers an intuitive, responsive interface for insights.

A social media automation platform built on the CrewAI Agentic AI framework, where multiple AI Agents collaborate to research trends, recommend niche topics, customize social posts, analyze performance, and suggest optimizations in real time. Leveraging RAG for contextual grounding and personalized content creation. Built with Next.js (frontend), Node.js API routes (backend), and Tailwind CSS for a responsive UI.

HckrNews

SEP 2025 – PRESENT

A full-stack Next.js tech news and discussion platform inspired by Y Combinator’s Hacker News, featuring hybrid search with Algolia and MongoDB, role-based authentication, story and comment threading, likes, bookmarks, reporting, and a full admin moderation dashboard. The platform enables real-time networking, project showcasing, and job postings for employers. It supports dynamic content rendering and API-powered real-time search, with Tailwind CSS for a responsive modern UI and a Node.js + MongoDB backend for secure, scalable data management.

Developed a feature-rich Java calendar application using strict Model–View–Controller (MVC) architecture and core SOLID principles (Single Responsibility, Open–Closed, Liskov Substitution, Interface Segregation, Dependency Inversion). Cleanly separated UI, control flow, and domain logic while supporting Swing GUI and command-line modes. Applied Strategy and Builder patterns for extensible, validated event creation, Factory pattern for multi-format exports, and Template Method for reusable controllers/views, with robust error handling and comprehensive unit testing.

Web-based application for customer purchase prediction using Multiple Linear Regression, leveraging product categories and user demographics to help clients target potential buyers with personalized discounts. Built with Python, Hadoop, and Azure ML Studio.