Velu Sankaran·Apr 9My AI Forgets Everything. So We Built a Wiki.The Karpathy LLM-wiki idea, applied to a content empire of 7 daily-publishing agents.
Velu Sankaran·Apr 2Hello, PyTorch! Let the Framework Do the CalculusSame models, same training loop — now with automatic gradients
Velu Sankaran·Apr 1The QA Platform That Writes, Records, and Heals Its Own TestsHow Checkmate evolved from an AI prototype into a full testing platform
Velu Sankaran·Mar 15From Line to Network: Build Your First Neural Network From ScratchA neuron is just y = mx + b with a bend — stack a few and you can learn curves
Velu Sankaran·Mar 15Hello, Model! Build Your First AI Model From ScratchNo frameworks, no magic — just Python, a line, and 7 housesA response icon1A response icon1
Velu Sankaran·Feb 5Test Your App Without Leaving Your EditorHow checkmate-mcp brings AI-powered testing to VS Code via the Model Context Protocol
Velu Sankaran·Jan 26Stop Writing Test Scripts, Start Describing TestsHow Checkmate turns plain English into running browser testsA response icon1A response icon1
Velu Sankaran·Jan 21UTML: A UI Test Markup Language for Browser AutomationIntroducing UTML — a JSON protocol for browser tests — and playwright-http, its reference executor
Velu Sankaran·Jan 11Your AI Agent Works in Dev. Why Does It Fail in Production?Unit tests, LLM-as-judge, A/B testing, and CI/CD pipelines for agents that actually work
Velu Sankaran·Dec 28, 2025My Coffee Went Cold. My App Was Done.Building a TP-Link smart switch controller with Claude Code in one sitting