Open-source work, AI tools, and engineering experiments.
CrewAI is designed to facilitate the collaboration of role-playing AI agents. In this example, these agents work together to identify & rank mutual funds based on the given inputs.
This project analyzes youtube talks using natural language processing, emotion tagging, storytelling classification, and cosine similarity for content clustering, the project identifies common narrative arcs, emotional triggers, and speaking strategies that consistently captivate audiences.
An enterprise-grade agentic AI platform that orchestrates a virtual team of specialised AI agents (MITRAs) to autonomously manage the entire software development lifecycle — from planning and design to coding, testing, and deployment. Governed by the SAARATHI console for human-on-the-loop oversight.
A read-only Model Context Protocol (MCP) server for querying LinearB data — deployments, teams, users, services, incidents, and metrics — via natural language through any MCP-compatible AI client like Claude.
A Streamlit app that mines storytelling patterns from TED Talks using YouTube transcripts. Uses NLP and optional GPT analysis to break down story structure — character, conflict, and emotional tone — to understand what makes a talk truly compelling.
Query ElasticSearch in plain English using a CrewAI multi-agent network. AI agents collaborate to translate natural language into ES DSL queries and fetch the required data — no query syntax knowledge needed.
A serverless AWS Lambda function (Python) that sets Slack statuses on a schedule — driven by a Google Sheet. Deploy once and let it run; no servers to manage.
A Flutter Web game where you and the computer each pick a secret number and take turns guessing. First to guess correctly wins. Features higher/lower visual clues, round history, and full keyboard support.