Why Local LLMs Don't Need C++ or Python: Building a 15MB Native AOT Inference Engine in .NET 10 September 13, 2026 · Dev.to Read full story at source