Enterprise AI & Workplace Automation KernelsSubmitted July 26, 2026

ModelSmithAI

An MCP app on the Model Context Protocol built by Kernels at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon and deployed on NitroStack.

About this project

ModelSmithAI is an autonomous, multi-agent system that turns a plain-English request into a trained, security-scanned machine-learning model in minutes. Six coordinated agents gather open-licensed images from the web, verify them, train a classifier, diagnose their own failures, gather better data, and deliver a model in the safe SafeTensors format with a PDF audit report. Built on NitroStack MCP, the same agents run behind a web app and an orchestratable AI chatbot. Who it's for: teams and individuals with no ML expertise, engineers facing niche tasks with no ready-made dataset, and MLOps/security teams who need to vet model files before running them. What's special: it's genuinely self-improving — the Diagnostician reasons about which classes fail and what data fixes them — and it treats security as a guarantee, converting every model to a format that physically cannot execute malicious code. The advantage: what normally takes an expert days or weeks, ModelSmithAI does in minutes — with no dataset, no labeling, and no code. Unlike AutoML tools that assume you already have data, it sources the data itself, improves through its own reasoning, and guarantees the final model is safe to use.

Enterprise AI & Workplace Automation track

Develop AI agents and automation tools that improve productivity, streamline workflows, and enhance business operations.

Source code

Team Kernels

  • Rohith Kumar S SLead

  • Nithish Raajan N K

  • Mohamed Navith H

  • Sanjay R.M.C

← All Amrita University Coimbatore projects

Frequently asked questions

What does ModelSmithAI do?
ModelSmithAI is an autonomous, multi-agent system that turns a plain-English request into a trained, security-scanned machine-learning model in minutes. Six coordinated agents gather open-licensed images from the web, verify them, train a classifier, diagnose their own failures, gather better data, and deliver a model in the safe SafeTensors format with a PDF audit report. Built on NitroStack MCP, the same agents run behind a web app and an orchestratable AI chatbot. Who it's for: teams and individuals with no ML expertise, engineers facing niche tasks with no ready-made dataset, and MLOps/security teams who need to vet model files before running them. What's special: it's genuinely self-improving — the Diagnostician reasons about which classes fail and what data fixes them — and it treats security as a guarantee, converting every model to a format that physically cannot execute malicious code. The advantage: what normally takes an expert days or weeks, ModelSmithAI does in minutes — with no dataset, no labeling, and no code. Unlike AutoML tools that assume you already have data, it sources the data itself, improves through its own reasoning, and guarantees the final model is safe to use.
Who built ModelSmithAI?
ModelSmithAI was built by team Kernels at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon, in the Enterprise AI & Workplace Automation track.
What is an MCP app and how is it built?
An MCP app is an application built on the Model Context Protocol — an open standard that lets AI agents connect to tools, data, and APIs. This project exposes MCP tools and resources that agentic AI systems can call. It was built and deployed on NitroStack, the full-stack platform for shipping MCP apps and servers.