AI agents are already involved in almost every stage of product development — from idea analysis to writing code. Does this mean the developer profession is losing its meaning? No — but the focus is shifting. In this talk, we’ll explore where agents are genuinely strong and where human judgment is still essential: architecture, scenario design, code review, and accountability for decisions. In the second part, we’ll run a live system design session on a real service (converting public transit schedules into GTFS format), walking through questions a developer should ask at every step — questions an agent won’t ask. We’ll see in practice the difference between "AI in the product" and "AI in the process," validate generated data against the official GTFS validator, and examine a case where an agent’s proposal had to be rejected to preserve the project’s architecture

We’ll talk about starting your career at Yandex. A recruiter will explain how the internship works, break down what to expect at each stage of the selection process, share preparation tips, and answer your questions

Textbooks teach us to choose algorithms by complexity, but in production services slow down more often because of data quietly moving around: a copy where a reference would do, a log message built only to be thrown away, a full rebuild when nothing has changed, a database replica added "just in case". All of this code is correct and passes the tests, and its price only shows up in the profiler or on the infrastructure bill. We'll go through real cases and learn to spot the cost of data in a function signature, a data structure and a database schema — before it reaches production

Caching sounds simple: remember a result and reuse it. Through real production examples, we’ll explore when caches save work, what they cost, and why a promising cache may not be worth building. Along the way, you’ll make a few engineering decisions and learn what to measure before adding your next cache

We will solve problems together in a mock interview. It’s a chance to get familiar with the technical interview format and receive feedback from an experienced Yandex interviewer

This talk explores dead code in the broadest sense—from unreachable branches and useless computations to unnecessary dependencies and outdated business logic that continues to consume CPU resources, increase mobile app size, and confuse both developers and AI tools. Using real-world examples, we’ll examine the consequences of dead code and outline a practical approach to dealing with it: how to identify candidates, prove they are no longer needed, remove them safely, and measure the impact after release

Your system has hidden weaknesses. The question is whether you find them before a real outage does

This talk is about using chaos engineering and controlled production drills to expose the things that normal testing often misses: hidden dependencies, critical services, fragile fallbacks, and unexpected failure paths

The goal is not to create chaos, but to run small, well-observed experiments, see how the system actually fails, and turn that into reliability work

A single user action can trigger a chain of service calls, database queries, data conversions, and retries. We’ll follow that journey to understand why correct code can become expensive at scale. We’ll explore database filtering, serialization and compression, caching, batching, retries, and load management. The goal is to learn to look beyond your own request handler: spot unnecessary work, know what to measure, and choose the right approach to each problem

Poor vehicle positioning is more than a misplaced dot: it distorts ETAs, complicates pickups, and undermines user trust. This lecture explains how map matching turns noisy GPS signals into plausible road positions—and how seemingly small errors propagate into real business impact. Finally, we put a new matcher to the test on real-world data—and discover why a trustworthy experiment is much harder than it looks