Bruno Capuano, conocido como “El Bruno”, es un Cloud Advocate en Microsoft que comparte en su blog guías y reflexiones sobre Azure, programación .NET, inteligencia artificial, y la intersección entre negocios y tecnología. En su sitio encontrarás artículos técnicos profundos, ejemplos de código en C#, análisis de rendimiento, demos de IA local y contenido dedicado a desarrolladores que desean llevar sus aplicaciones al siguiente nivel. Su estilo es directo, con voz experta y orientado a solucionar problemas reales en entornos
80 articles from this blog
A developer deploys a C# agent to Microsoft Foundry but discovers it responds as ChatGPT, tracing the issue to deploying from the wrong Git revision.
Learn how to use the ElBruno.S1Mini NuGet package for local ASR transcript normalization in .NET using the s1-mini ONNX model.
A technical guide on connecting GitHub Copilot to Microsoft Foundry models using a custom Azure OpenAI provider and BYOK setup.
A blog post announcing a 4-part Microsoft Reactor series for C#/.NET devs to build an AI agent harness, live coding a 'claw' agent with tools, memory, and deployment.
MagenticBrain model now supported in ElBruno.LocalLLMs for local .NET agentic workflows with ONNX package.
Fara 1.5-9B model now fully supported in ElBruno.LocalLLMs for C# local inference with vision capabilities.
Explains why a 975B-parameter MoE multimodal model can't run locally in .NET via ONNX Runtime GenAI due to size, architecture, and conversion limits.
Building a weekly AI news digest using GitHub Agentic Workflows, with automated curation and pull request review.
A .NET global tool that monitors all Foundry Local AI model instances on your machine, showing loaded models and device backends in real time.
Learn how to run local AI models for Copilot tasks using Ollama, Foundry, and Azure OpenAI proxies in VS Code.
A C# library bridging Foundry Local SDK with Microsoft.Extensions.AI and Agent Framework for local-first AI agents.
Tips on optimizing GitHub Copilot token usage to avoid high costs, with practical advice for developers.
Adding MAI-Image-2.5 support to ElBruno.Text2Image .NET library and t2i CLI for image generation from terminal or C# code.
VS Code 1.122 enables BYOK models without GitHub sign-in, simplifying enterprise and offline AI workflows.
A reference app combining NVIDIA NeMo, Microsoft Agent Framework, A2A, and Aspire for cross-framework multi-agent workflows.
Experiment comparing GitHub Copilot CLI with SQUAD and GPT-5.5 BYOK for app building, highlighting better AI but persistent need for engineering discipline.
Experiment using GitHub Copilot CLI with GPT-5-mini via BYOK to build a Windows tray app, highlighting that code generation is cheap but quality gates are expensive.
Developer shares how GitHub Copilot helped transform pet projects into over 20 open source NuGet packages for AI, embeddings, and productivity.
Running GitHub Copilot CLI offline with GPU-accelerated local models, comparing performance with CPU-only setup in a real .NET project.
Testing GitHub Copilot CLI offline with local AI models on CPU-only hardware, evaluating performance for .NET development tasks.
