I'm Mos Ahmed. I write about building AI systems that actually work — from Claude Code implementations to production architectures. No hype, just honest engineering notes.
A computational algorithm that uses random sampling to solve problems that might be deterministic in principle. Like throwing darts at a board to estimate π! Powers AlphaGo, reinforcement learning, and MCTS.
An Anthropic engineer spent $9 on a one-shot coding agent and got broken code. The same team rebuilt it with a proper harness for $200 and shipped a full-stack app. A field guide to the four failure modes and five primitives that are quietly winning the agent race in 2026.
A first-principles breakdown of how LLMs produce deterministic probabilities but non-deterministic outputs, and practical strategies to control randomness in Claude Code.
A complete guide to implementing, deploying, and monitoring AI agents using Claude's latest capabilities. Includes cost analysis, error handling, and real-world performance metrics from production deployments.
Learn to create reusable skills that extend Claude Code's capabilities. From SKILL.md format and resource files to discovery and best practices for building powerful custom workflows.