rigout
PackageA Python package: an MCP server that lets AI agents use a real machine — run commands, manage files, use Docker, keep terminal sessions alive. pip install rigout.
Install it on a machine, hand the printed URL to an AI agent, and the agent can work there: commands, files, Docker, persistent terminals. Auth on by default in public mode. My first Python package, live on PyPI.
Browser tools and open-source packages, built with AI agents and curated by hand
A Python package: an MCP server that lets AI agents use a real machine — run commands, manage files, use Docker, keep terminal sessions alive. pip install rigout.
Document Q&A you can inspect: hybrid retrieval, OCR ingestion, and page-level citations with every answer. FastAPI, Qdrant, Docker.
Ask questions about your own documents and get answers with citations.
Record and manage voice notes in your browser.
View and interact with 3D models directly in your browser.
Transcribe speech to text in your browser, in real time.
AI-generated video experiments, published on YouTube.
Record audio and video from your browser.
Zero-shot image classification in the browser: label images without training a model.
Browser-based audio editor with marker-based selection, effects, and WAV export.
Peer-to-peer voice calls in the browser, built on WebRTC.
Turn photos and videos into line-art animation, processed in your browser with no uploads.
A curated library of prompts and instructions for Jules and other AI coding agents.
Send data between devices using sound — information encoded into audio, with send and receive modes.
I'm Melbin J Paulose. Everything on this site was built by directing AI coding agents: I define the problem, set the boundaries and what a tool must never do, and validate the result by using it. My strength is taking an idea to a working product and iteratively improving it.
I'm currently open to junior AI engineering work — AI automation, RAG workflows, and agent tooling.