Schedule

In theory, database sharding is a relatively simple concept. In practice, however, things get much more complicated.

In this talk, we will discuss possible approaches to database sharding and look at the reasons why implementing this technique may potentially break your production code.

We will discuss how to implement delivery guarantees, including exactly-once delivery, and what obstacles there are.

Let's look at proven approaches, optimization methods, and non-obvious and insidious features of PostgreSQL.

During the master class we will together define what variants of cutting a monolith can be, break an existing monolith into limited contexts, and design several steps for taking parts out of a monolith. We'll also invent some patterns and discuss the pros and cons of the implementations.

Here is an example of a task from a production project for which DomainEvents will be useful. I'll show you the difference between DomainEvents and ApplicationEvents. I will share the current implementation of DomainEvents on a modern stack.

During the master class we will together define what variants of cutting a monolith can be, break an existing monolith into limited contexts, and design several steps for taking parts out of a monolith. We'll also invent some patterns and discuss the pros and cons of the implementations.

Architecture design and validation are becoming the main field of activity in an age where code is left to AI. Let's talk about automating architecture quality control and an open-source tool that makes it easy to implement.

Microservices offer independence, but at the cost of overhead, harder debugging, and complex deployments. The alternative is a modular monolith: one application, one process, yet the code is divided into independent modules with well-defined boundaries.

In this session, I'll demonstrate how to build this architecture in vanilla ASP.NET Core — using HostingStartup, ApplicationParts, and EF Core's IEntityTypeConfiguration. No external libraries, no rewriting existing code. And it all runs in a single container and starts with F5.

We will discuss how to implement delivery guarantees, including exactly-once delivery, and what obstacles there are.

At the Event Storming workshop, you will learn how to effectively use this powerful tool for system design.

Market research overview: DORA, bigtech reports, and our own AI4SDLC 2026 research.

You will learn how to get predictable, controlled, observable, secure, if-you-have-managed-to-read-here-is-the-adjective-out-of-bullshit-bingo AI development with an AI team.

To stop agents from getting stuck at "here are the changes, take it from here," they need a harness — an engineering loop of project memory, skills, and MCP. Drawing on my experience at Yandex, I'll show how to build that environment and how the engineer's role changes along the way.

A story about a home AI lab.

Microsoft Agent Framework is a dark horse. It's clear that it's for writing agents, but it's not clear why. We had a task that this technology was perfect for. I will discuss how MAF works and how we used it in my talk.

How we have built a production-ready AI assistant that understands team context and works seamlessly with .NET infrastructure at SoftLight.

How to create a control system for an AI agent using the .NET ecosystem.

Debunking the SIMD myth in .NET as a "magic button" for guaranteed acceleration, and using a real case of matrix calculations, we show why vector code can run idle.

We will take the same code examples and compare them using different parameters.

We'll learn the differences between EventSource, DiagnosticSource, and ActivitySource. We'll learn how to implement them in our own and other people's code.

Let's discuss what discriminated unions are, how they work, and what prospects they open up for developers.

In this talk, I will compare the performance of Native AOT and JIT in .NET 10 on x86-64 and ARM64 processors using real-world algorithms, and demonstrate how significantly the choice of processor instruction sets (SSE/AVX2/AVX512) affects the speed of AOT-compiled applications.

I will also cover undocumented project settings and proper benchmarking of Native AOT applications with BenchmarkDotNet. Additionally, I will discuss the upcoming changes in code generation planned for .NET 11 and their impact on performance and processor compatibility.

C# 14 Extension Members let APIs defined outside a type appear as part of that type’s own API: extensions can now include properties, static members, and operators. In this talk, we’ll explore how this expressiveness affects .NET API design, compatibility, and context-dependent behavior — and where convenient syntax helps, where it hides business logic, and where it introduces accidental complexity.

A real story of how my team went from 30—40 minutes per integration test run down to 3—4 minutes. We'll talk about how we got there, the mistakes we made along the way, and where we ended up.

I'll show you examples for both EF Core and LINQ to DB in .NET.

Let's prove that TDD is a superpower, but it's not rocket science and is accessible to everyone.

We have launched an experiment in Kontur to create our own IDE. In the talk, I will tell you what came of it.

Let's discuss why, despite the availability of heavy-lift launch vehicles, there are still no large-scale data centers in space, whether they are actually needed, and what the supporting infrastructure for computing in space might look like.

We'll also try to figure out how to write and deploy .NET code on a real spacecraft, while exploring Microsoft's history in the space industry — how the company entered the field, where it stepped back, and which of its technologies and contributions are still in use today.

We'll be opening the conference for .NET Developers — DotNext 2025.

The Ozon team invites you to play bingo with logical and mathematical puzzles — at first glance, they seem simple, but each one has a hidden twist that engineers will appreciate

Try yourself as a speaker and tell us about everything you care about right at the conference.

A talk on how any complex system can now be wrapped in an interface layer that will provide any user — even those who aren’t technically inclined — with a concise and secure way to send requests and commands in a more "human‑friendly" style.

We'll be summarizing the results of the DotNext 2025 conference.