An AI-native Markdown indexing system and MCP server that structurally maps documents into section trees, allowing LLM agents to fetch specific contexts instead of consuming entire files, drastically reducing token usage and improving accuracy.
Distributed locking library on top of RDBMS, published to npm. Exposes Spring Boot auto-configuration so teams can drop it in without boilerplate.
A local LLM orchestration system where specialized models take on distinct roles — each validating the other’s output — to produce results rivaling frontier models like Claude Opus, without sending data to the cloud.
Building the Pixel Credit Card.
This blog explores subtle yet dangerous behaviors in C++ and Java that can silently lead to bugs despite valid-looking code. It breaks down why these issues exist and how language design choices trade safety for performance and flexibility.
This blog breaks down the reality behind Claude AI building a C compiler, separating the hype from the actual technical limitations. It explores what the achievement truly means for software engineers, where AI still falls short, and why human architects remain irreplaceable.
Ngrok tunnling allows you expose your local hosted apps to the internet.
