Cloud as a Story - Vunvulea Radu

A few days ago, I listened to one of The Cloud Pod's podcasts, and they mentioned Microsoft's approach regarding Copilot and how Microsoft does not have its own LLM model. I started to dig a little deeper into this topic, as it has high potential regarding training the team to use one 'interface' and behind the scenes to be capable of switching between different LLMs. Many think that 'Microsoft Copilot = one big model from OpenAI'. This was true initially, but today Copilot is more like an air‑traffic controller for AI. It can work with several large language models (LLMs), route your prompt to the right place, and bring back an answer grounded in your work data from Microsoft 365. The important part is which model and how the Copilot system chooses actions and protects your data. What multi‑LLM Copilot means Inside Microsoft 365, Copilot sits on top of an orchestration layer. This orchestrator is the interface between foundation LLMs and the skills and actio...

  I like to think about this trio (Practice, Capabilities, Communities) like a query: precise inputs, clear joins, and an output that matters to the people & business. SELECT Practice is the formal expression of what we sell. It lives at the intersection of commercial need, company offers (services), and the capabilities we've matured because the market or our strategy demands them. When a client asks for Cloud Migration, AI Enablement, or FinOps at scale, that's the Practice answering. Capability is how the work gets done. It's the skill, tooling, and method layer—architecture patterns, IaC, MLOps, data engineering, and SRE working methods. Capabilities power one or many Practices. Community is the human engine, the real force behind the numbers. It forms around shared interests—people who want to learn, try, and trade notes. Communities are where ideas incubate and capabilities grow long before they're ready to be packaged into a Practice. FROM Practices draw fro...

A few years ago, I would have said monitoring was good enough for most systems. But then cloud systems exploded in complexity, and everything changed. These days, when dealing with loads of microservices, containers, and distributed applications running all at once, I quickly realised that monitoring alone just points out that something’s broken. It rarely tells the whole story. Observability is what’s changed the game for me and many engineers. It’s about seeing under the surface, connecting the scattered pieces — metrics, logs, traces, dependencies — and understanding how everything interacts. It’s the difference between knowing there’s a fire alarm going off and knowing where the fire is and what caused it. I’ve seen firsthand how missing observability leads to playing whack-a-mole with incidents. It becomes a constant scramble to fix things after the fact. According to a 2024 Grafana Labs survey, over 80% of organisations still struggle to trace incidents to the root cause in t...

 When we talk about cloud integration, many people think first about connecting the APIs. But the real challenge comes when we talk about scaling. How do we ensure that a system continues to work when traffic grows, when events come faster, or when one side of the system scales differently from the other? In this post, I want to show how we can connect Azure Event Grid, Azure Functions, and AWS Kinesis in a way that scales correctly. The idea is simple — we read events from Event Grid, transform them in an Azure Function, and push them into Kinesis — but making it scale well across clouds needs some attention. Azure Event Grid — it’s all about scalability Many people believe that Azure Event Grid scales automatically without limits. That is not really true. Event Grid can handle many events , but has its own capacity model. Creating a custom topic in Event Grid is internally partitioned to allow more parallel processing. Still, there are hard limits per topic. For example: ...

When we talk about Artificial Intelligence today, it feels like a revolution. Everyone wants to try it, but the reality is hard: studies show that up to 80% of AI projects never reach production. Often, the reason is not the model itself but the team behind it. Too many companies start with a clever prototype and then get stuck, unable to deploy or monitor at scale. This is where Microsoft Azure, together with the right team shape, makes the difference. More Than a Model: A Full Cloud AI Solution An AI solution in the cloud is never just a model. It is a complete ecosystem of applications, infrastructure, data pipelines, and security. On Azure, we can connect all of this. Azure Machine Learning gives us model lifecycle management, Azure OpenAI Service brings natural language power, and Cognitive Services add vision or speech. This runs on a secure, automated, and scalable cloud infrastructure. For business leaders, this means faster time to market and real outcomes, not just pilots. Fo...

  Cluj just hosted DevTalks , one of the biggest and most vibrant tech events in Transylvania. The event brought together thousands of developers, engineers, and innovators under one roof, and the energy was fantastic. I had the chance to take the stage and talk about something very close to my heart: how to build and structure teams for Cloud AI projects. We often talk about the technology — the models, the platforms, the automation — but what really makes or breaks success is the team. My message was simple: building AI in the cloud isn’t a one-person show. It takes AI/ML experts, data scientists, cloud and platform engineers, developers, and security specialists all working side by side. The real magic happens when you get the balance right and put automation and platform engineering at the core If I had to leave the audience with one idea, it’s this: “AI in the cloud is a team sport. When you get the structure right, innovation, scalability, and cost efficiency follow natur...

In March 2025, I had the opportunity to be a speaker at the Conf42 conference. The session that I delivered covered the cloud architecture that needs to be build to manage the payload generated by AI.   Refs: https://www.conf42.com/Cloud_Native_2025_Radu_Vunvulea_data_infrastructure_algorithms