Nabaraj Ghimire

I delivered a session titled “AI-ready SQL MCP Server with .NET” at an in-person Microsoft Build 2026 community event held on June 20, 2026, at YCOTEK, Lalitpur, Kathmandu. The event was focused on helping developers and cloud engineers design, build, and deploy real-world AI solutions using Microsoft technologies such as Azure AI, Microsoft Foundry, and GitHub Copilot. In my session, I shared how developers can build an SQL MCP Server using .NET to connect AI-assisted workflows with database systems in a safer and more structured way. I demonstrated how MCP can help expose database capabilities with controlled access, configurable permissions, and practical integration patterns for AI-native applications.

I recently published an article explaining how to automate the deployment of an ASP.NET Core 9.0 application to a remote Windows Server running IIS using PowerShell. The article covers a complete deployment workflow, including building and publishing the application, securely handling credentials through Windows Credential Manager, stopping the IIS application pool, showing an `app_offline.htm` maintenance message, removing old files, copying new files, restarting the application pool, and opening the deployed website for verification. The goal of the article was to help developers reduce manual deployment effort, avoid common deployment mistakes, and create a reusable automation approach for staging or production environments.

I delivered a session titled “AI Native Data Application with SQL MCP” at Microsoft Build 2026 // localhost: Kathmandu, a local community event organized to bring the Microsoft Build experience to Nepal. In my session, I shared how SQL MCP can be used to build AI-native data applications by connecting AI-assisted workflows with database systems in a more structured and practical way. I explained the role of MCP in enabling safer and more useful interaction between AI tools and data sources, and discussed how developers can apply this concept in real-world software development.

I spoke at a community-led developer event on GitHub Copilot held at Presidential Graduate School, Kathmandu. My session, “GitHub Copilot: Your AI Companion for Every Workflow,” focused on how developers can use Copilot in everyday work, not only for code generation but also for understanding existing code, refactoring, debugging, documentation, and improving productivity. I also demonstrated practical examples so attendees could see how Copilot fits into real development workflows. The goal was to help local developers become more confident with AI-assisted coding and encourage practical adoption of GitHub Copilot in the community.

The blog explains TOON as a token-efficient format for structured LLM prompts. TOON helps to reduce JSON clutter like braces, quotes, and repeated keys. Its impact is to lower token usage, reduced cost, faster responses, and better context use in agents, copilots, and RAG pipelines.