Sakari Nahi

CEO at Zure, Microsoft RD, AI & Azure MVP

About

Sakari is the other founder and CEO of Zure, an Azure-focused company. He has been coding on .NET professionally since 2002. He's a dad of two, husband of one and likes roguelikes & text-based games. He's passionate about emerging technologies, distributed computing, and synergies thereof. Sakari is also a Microsoft Regional Director, and AI+Azure MVP. He organizes Finland Azure User Group and casts a pod Ikkunastudio in Finnish. Zure designs, develops, and manages solutions on Azure, with offices in Finland, Belgium, Denmark, United Kingdom and Netherlands.

Book This Speaker

Sessions Available

Multiagentic app architecture

In the last couple of years, development of AI applications has matured. As an industry, we have found some design patterns, and what's more, enterprises and business decision-makers have figured out sensible business use cases for LLMs. I've been part of over 30 GenAI cases during the last couple of years, and in this talk I will open the current state of AI application architectures, be they single or multi-agent. I will also look at what will happen next regarding ourselves (the technical designers and implementers) and organizations (who struggle with the governance, adoption and regulations of new technologies). After this talk, you will have an understanding of a multiagentic AI app architectures, why are they needed, and what could happen next the evolution of AI app architectures.

State of GenAI-based development in Azure

GenAI is the largest hype. All products have AI, and all organizations are interested in it, with the highest levels of leadership demanding AI capabilities from the IT and development teams. Yet, at the same time, there's only a limited number of meaningful software development projects with GenAI. Why is this? What is going on? What is going to happen next? Where will we be next year? This presentation explores the current state of GenAI development market, followed by practical AI projects and forecasting for what's next regarding this highly hyped and interesting field of GenAI.

Modernization of HUGE legacy apps in Azure

We've all been there. There's a legacy codebase that's old enough to go to a bar, and "surprisingly" it has become such a big ball of mud that adding new features or fixing bugs has become very time-consuming and expensive. And irritating enough for some people to consider resigning. At last the leadership team makes the decision - this big ball of mud needs to be modernized! Let's cloudify it, how hard can it be? The goal is to make the app more maintainable and easier to develop, as well as increase the user happiness. Maybe lower the runnings costs too! App modernization, meaning "modifying solutions to fit cloud", has been and will continue to be one of the hot topics in any company's cloud strategy. Unfortunately, modernization projects tend to fail at least partially. This talk explores that failure. I have over 2 decades of experience from this field, and as a CEO of a development company, I get to talk about these issues a lot with leadership and development teams. This talk is aimed at decision-makers, developers, team leaders, and all stakeholders who are part of a modernization project. After this talk, you'll know what to do to make your modernization project succeed.

Please, don't do microservices

Most big thinkers in our field agree: microservice architecture is rarely the correct choice. For 1/100 of projects, there might be a path forward with microservice architecture. Do you think your project is special enough to be that one? If you do, please attend this session. In this session, not only will you come to understand microservices, you will also learn the pitfalls of the architecture as well as how to mitigate the risks and approach the creation of a microservice architecture. After this session, you will be able to identify the situation where microservice architecture is a viable choice to go forward. In addition, you will know how to actually go forward. The goal is that after this session, many of you will view microservices differently.

Real-time design of an Azure PaaS app - Improv

In this session, you will see how a person with over 10 years of Azure PaaS and 20 years of software development experience designs a solution with Azure PaaS components. The promise is, you will have a broader view and understanding on the components and approaches used. The session consists of 10 minutes of PaaS introduction, and the rest of the time of real-time design of an application with draw.io on the screen. The idea for the application is received from the audience, and nothing is decided before the session. The presenter talks their thinking and decision-making out loud, allowing you to follow his thinking his decision-making of Azure services. This session is most applicable for developers and architects, but is followable for other roles, such as IT professionals and decision-makers, and can indeed give them a glimpse into what does it mean to be an developer/architect. The session is interactive, and audience is invited to share their opinion and extend the required features. The finalized design diagram is shared afterwards, including a description of core features. The session is about level 200.

Modern Azure Data Platform

After this session, you will be familiar with the positioning, responsibilities, and relevant Azure tools for a modern, Azure-based data platform. Data platform consists of several responsibility areas, such as integrations / IPaaS, APIs / API Management, ETL and IoT streaming pipelines, machine learning, and, of course, data warehousing and BI. In this talk, the presenter will present all of these areas and their responsibilities, including the relevant and most commonly used Azure services, with a decision-tree on what services to pick when. This session is most applicable for enterprise and solution architects, data experts (engineers/scientists), and developers, but can also be mostly followed by IT professionals and decision-makers wishing to understand the cloud stack a bit better. The session is about level 250.

Selecting the right Azure products for your Azure PaaS project

Microsoft Azure just keeps on growing. The amount of new releases during the year is alarming. In addition, many of the new products overlap with the old ones or with each other. This talk gives you a quick overview of the meaningful differences between the Azure products commonly used when building a custom solution on top of Azure PaaS. The commonly used product consist of Azure App Services, Azure Functions, Azure SQL, Azure Cosmos DB, Azure Search, Azure Data Lake, Azure Redis Cache, Azure Service Bus, Azure Event Grid, Azure AD, Azure API Management and Application Insights. The talk is aimed at beginner and intermediate levels and those of us who have not had the time to broadly explore Microsoft Azure, with some bits at advanced level.

Disruption is the job

15 years ago I was a coder, but as my friend and I founded Zure, I became a CEO. This talk is about my views as a CEO of an independent company in 5 European countries. Our industry is facing a disruption in the form of AI. We have faced other disruptions earlier, such as Internet, Mobile, Cloud, and others. Are disruptions a threat, or an opportunity? The aim of this talk is to lessen your anxiety and increase your excitement, and if you happen to have FOMO or imposter syndrome, hopefully you'll have less of them after this talk.

Improv: Let's design an Azure AI app together

In this session, we will design an Azure AI app live. The idea for an app can come from the audience. The requirement is that the idea has features that can be implemented with generative AI capabilities. In the session, I will walk through my thinking on the topic while drawing the architecture together with the audience. This session is most applicable for developers and architects, but is followable for other roles, such as IT professionals and decision-makers, and can indeed give them a glimpse into what does it mean to be an developer/architect. The session is interactive, and audience is invited to share their opinion and extend the required features. At the end of the session, you will have an understanding of relevant Azure services as well as on the process of an architect in how they approach designing an app.

Spec-driven development live

In this session, we will SDD something together. I'll ask you what, and then we will explore different ways of approaching vibecoding and SDD. The goal is to get as close as possible to proper coding. At the end of the session, we might not have something that compiles, but we will certainly have learned something.

From Vibecoding to Team Coding

Many of us developers have tried vibecoding or spec-driven development. Some of us think it is a real delivery boost, others that the quality is not there yet. Some have tried SDD frameworks, such as Spec-Kit or Kiro, while others play around with simpler PRD.md files. Some swear on the process of research-plan-implement, with a sprinkling of git work-trees and parallelization via subagents, while others think that AI is more limited and best suited to learning, documenting, planning and maybe generating tests. I don't know who is right, but I know that in a team of people, not everyone can have a different approach. This talk is about my experiences in how a team could approach AI-assisted development. After this talk, you'll have the necessary tools to start figuring out if AI is something that might boost your team's delivery. The talk is not technical per se, more about the tooling, process, and frameworks such as SDD, context, and harness engineering.

From Single Agents to Systems: Designing Multi-Agent Architecture

In the last couple of years, architecture of AI solutions has matured. As an industry, we have found some design patterns and best practices as we've realized that single-agent designs may struggle with reliability and scale when we move from experimentation to production. In this talk, we explore how multi-agent architectures address requirements by decomposing AI systems into specialized, coordinated components, with an understanding of the trade-offs. After this talk, you will have an understanding of the different approaches to multi-agent architectures, why are they needed, and what could happen next in the evolution of agentic development.