GitHub Copilot Dev Days × Global Azure 2026 — Mumbai
AI can help you build faster — but a great app still needs a great home. ☁️
The wait is over! After an incredible response to our Call for Papers, the agenda is officially locked, and registrations are now open.
This edition of GitHub Copilot Dev Days in Mumbai is a focused, hands‑on community meetup centered on AI‑assisted development with GitHub Copilot, complemented by two Global Azure sessions that explore how modern applications move from code to cloud.
The event is primarily about shipping better code with Copilot — and then taking the next logical step: running, scaling, and operating those applications on Azure.
🚀 The Agenda
Track 1: GitHub Copilot Dev Days
Focusing on Spec-Driven Development, AI Agents, and advanced developer productivity.
9.40 am - 10.00 am
Session 1: GitHub Copilot: Your AI Companion for Every Workflow
Speaker: Augustine Correa
Bio: Microsoft MVP for Developer Technologies and AI Platform
10.00.00 am - 10.30 am
Session 2: From Developer to AI Architect: Beyond Autocomplete with GitHub Copilot
Software development is no longer about writing every line of code. It's about orchestrating intelligence.
In this talk, we’ll explore how modern engineers can evolve into AI-powered architects by leveraging agentic workflows with GitHub Copilot. Instead of using AI as a simple autocomplete tool, we’ll demonstrate how to use Copilot’s Agent Mode, CLI integrations, and automation capabilities to handle complex tasks like multi-file refactoring, system redesign, and DevOps workflows.
Through real-world examples including scaling a backend system and automating cloud workflows, you’ll learn practical mental models to integrate AI into your daily development lifecycle.
By the end of this session, you won’t just “use AI” but also, you’ll delegate, orchestrate, and build with it.
Speaker: Mubarra Kausar
Mubarra Shaikh is a backend engineer focused on building scalable systems and exploring how AI is transforming modern software development. With experience in Python and frameworks like Django and FastAPI, she has developed end-to-end applications including real-world systems in domains such as healthcare and hospitality.
Her work centers on AI-assisted engineering, where she uses tools like GitHub Copilot to refactor legacy code, design clean architectures, and accelerate development workflows. She is particularly interested in agentic workflows, where developers move from writing code to orchestrating intelligent systems.
10.30 am - 10.50 am
Session 3: SDD in the Trenches: Taming Copilot Agents in Community Codebases
As GitHub Copilot agents become part of everyday development, a new bottleneck has emerged: unclear intent. When Copilot is driven by vague or non‑deterministic prompts, it generates code that is inconsistent, brittle, and misaligned with project architecture. This problem is amplified in community‑driven FOSS projects, where architectural rules are often implicit or undocumented.
In this session, we share lessons from building and maintaining Visage—our primary Spec Kit–driven platform—along with hackmum.in, our community website, by adopting Spec‑Driven Development (SDD) as the control layer for Copilot agents. Using Spec Kit and the lightweight OpenSpec framework, we show how specifications become the primary instruction set for AI—before Copilot writes a single line of code.
Instead of prompt‑first development, we demonstrate spec‑first, agent‑executed workflows that turn Copilot from a guessing engine into a predictable collaborator.
What you’ll see
Takeaways
Attendees will leave with an open‑source blueprint for using GitHub Copilot agents responsibly:
Improve determinism in AI‑assisted coding
Preserve architectural intent in community projects
Scale contributions without increasing review burden
This talk is for developers who want Copilot agents to amplify engineering discipline - not replace it.
10.50 am - 11.30 pm
Session 3: Breaking and Securing MCP Servers: Real‑World Risks and Defenses
The Model Context Protocol (MCP) is quickly emerging as a standard way to connect AI agents with tools, services, and data sources. As adoption grows, one critical question remains: how secure are MCP‑based systems in practice?
In this demo‑heavy, security‑focused session, we will explore real‑world risks in MCP implementations by demonstrating how poorly protected MCP servers can be compromised. Using controlled, educational examples, the session will showcase attack vectors such as:
Prompt injection through poisoned or untrusted data sources
Tool poisoning and tool shadowing attacks that manipulate agent behavior
From there, we’ll shift focus to defense. Attendees will learn practical mitigation strategies and review a secure Azure reference architecture designed to reduce blast radius, enforce isolation, and harden MCP deployments.
We’ll also examine:
1. How Docker Sandbox isolation features can be used to limit agent execution risk?
2. How these sandboxing approaches integrate effectively with the GitHub Copilot CLI for safer AI‑assisted workflows?
Attendees will leave with concrete, actionable guidance on identifying MCP risks, securing agent‑tool interactions, and building more trustworthy AI systems: grounded in real-world demos rather than theory.
Speaker: Maninderjit Bindra
Maninderjit (Mani) Bindra is a Principal Software Engineer at Microsoft, focused on AI security and cloud security. He is a maintainer of the open-source Minimum Permissions Finder (MPF) project and an active contributor to open-source communities. Check out his OSS contributions at https://github.com/maniSbindra and blog at https://medium.com/@maninder.bindra
Global Azure Interlude [while we change the setup for the GitHub Copilot workshop]:
Session 1: From Root to Rootless: Securing Containers One Layer at a Time
"Containers feel safe but hide dangerous defaults. Through live demos, we'll exploit a vulnerable app running as root, then progressively lock it down — non-root users, distroless images, rootless containers, and finally microVM sandboxes. Each demo builds on the last, showing exactly what each security layer prevents and why it matters for your production workloads.
Speaker: Maninderjit Bindra
Maninderjit (Mani) Bindra is a Principal Software Engineer at Microsoft, focused on AI security and cloud security. He is a maintainer of the open-source Minimum Permissions Finder (MPF) project and an active contributor to open-source communities. Check out his OSS contributions at https://github.com/maniSbindra and blog at https://medium.com/@maninder.bindra
Session 2: AI Agents in Production: Controlling AI Cost and Traffic with Azure APIM
One AI agent serving many tenants can get expensive and hard to manage very quickly. In this session, I will show a simple Azure-based setup where Azure API Management helps control traffic plus costs, protect backend APIs, and reduce unnecessary AI usage. We will look at practical ideas like tenant-based rate limits, basic caching, and safer API access.
Speaker: Hardik Mistry
Hardik Mistry is a Principal Architect and hands-on cloud engineer with experience building scalable SaaS systems across Azure, .NET, APIs, distributed systems, and platform architecture. His work focuses on turning practical product and infrastructure challenges into reliable, production-ready solutions. He is especially interested in cloud-native application design, multi-tenant systems, and how AI agents can be deployed safely and efficiently in real-world environments. Alongside engineering, he also shares insights on AI, cloud, and modern software architecture through technical content and community-driven discussions.
Workshop + Lunch + Surprise Project Debut
Build a Social Bingo game while mastering VS Code's Agent Mode with GitHub Copilot. Learn context engineering, custom agents, and AI-first development workflows.