Engineering Reliable AI Systems on Anthropic Claude & Open-Source Rust Infrastructure.
EzLabs builds production-grade foundation software for autonomous AI agents and high-throughput LLM workloads.
A lightweight, compiled reverse proxy built in Rust for resilient LLM routing. Features zero-copy streaming, multi-account quota pooling, Token Saver prompt compression, and fallback routing across Anthropic Claude, OpenAI, and local endpoints.
An experimental research direction exploring Model Context Protocol (MCP) execution harnesses for scoped agent coordination, capability boundaries, and state checkpointing.
An internal evaluation toolkit planned to fuzz agent outputs against strict JSON schemas and verify behavior against safety guidelines before production deployment.
Founded by Nam Hoang (Lead Systems Architect), with core systems engineering by Huy Hoang and client architecture by Khoa Nguyen.