The latest articles on DEV Community by Sutaniese (@sutaniese). https://dev.to/sutaniese https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3824186%2Fb720ebd1-8a96-4a01-ac44-22a9f070be14.webp https://dev.to/sutaniese en Sutaniese Sat, 14 Mar 2026 16:29:39 +0000 https://dev.to/sutaniese/im-15-based-in-kazakhstan-and-i-built-an-mcp-server-for-ai-agents-to-handle-ml-datasets-33cp https://dev.to/sutaniese/im-15-based-in-kazakhstan-and-i-built-an-mcp-server-for-ai-agents-to-handle-ml-datasets-33cp <p><a href="https://getvesper.dev/" rel="noopener noreferrer"></a></p> <p>I'm 15 and based in Kazakhstan. I started coding seriously about a year ago. No CS degree, no team, just figuring things out.<br> I got obsessed with AI agents - specifically why they're so capable at reasoning but completely fall apart the moment they need real data. Every pipeline I tried to build had the same bottleneck: the agent couldn't search for datasets, evaluate which ones were actually useful, clean them, or export them. All of that still needed a human.<br> That felt like a solvable problem. So I built Vesper - an MCP server that gives AI agents the full ML dataset workflow. Search, download, quality analysis, cleaning, export. Fully autonomous.<br> I'm still in school. Built this between classes and after homework. It's live, has real users.<br> Early stage, brutal feedback welcome - getvesper.dev or try it directly: npx vesper-wizard@latest</p> mcp devtool machinelearning ai