X5 Tech added AI skills assessment to developer interviews

X5 Tech has integrated AI skill verification into technical interviews and stopped pretending that developers work without such tools. The company wants to see not a sterile solution to a problem, but a real process: how the candidate uses a model, verifies it, and takes responsibility for the result.

How X5 is changing interviews

A pilot has already been launched for senior Python developers and QA engineers. At the screening stage, candidates are warned that there will be a separate AI block in the interview. The company provides the tool itself, so the person doesn't need to bring their own stack or subscriptions. During the session, the candidate shares their screen and solves the task together with the model. For interviewers, what matters more than the final answer is the trajectory of thinking: how quickly the person formulates a prompt, clarifies the initial conditions, notices weak spots, and changes approach if the model goes off track.

The logic is simple: if in everyday work an engineer uses AI anyway, it's pointless to prohibit this during an interview. X5 Tech directly states that they don't expect the image of a "developer from the past" who keeps everything in their head and principally avoids assistants. At the same time, lack of extensive experience with AI is not considered an automatic minus. If the candidate is strong in basic engineering, they're ready to hire them and improve their skills internally through tools, guides, and in-house courses.

What now counts as a strong signal

The interview focuses not on the fact of using AI, but on the quality of work with it. A candidate may be asked what tools they apply, what they understand by context window, tokens, model temperature, agentic approach, and MCP. Another type of task is to give a file like CLAUDE.md or .cursorrules and ask them to evaluate whether it would help an agent, where there are ambiguities, excess context, and potential hallucination triggers. In other words, they check not memory, but engineering thinking around AI.

• what AI tools a person actually uses and in what scenarios
• whether they can write prompts with role, context, and expected answer format
• whether they understand model limitations, including hallucinations and context size
• whether they notice security risks and don't send secrets to public services
• whether they can critically review AI-generated code rather than just inserting it into the project

For juniors, the bar is still basic: they need to understand the fundamentals of prompt engineering, be able to recognize hallucinations, and not confuse quickly generated code with real understanding of the system. In their article, X5 Tech specifically emphasizes the problem of AI dependency among junior developers: they assemble solutions faster, but worse at explaining why it works, how to debug it, and where its limits are. That's why a good signal becomes not a set of random pet projects, but several carefully documented works with a README, the author's role, decisions made, and explanations of where the model's advice had to be rejected.

"Whoever merged the code is responsible for it."

For mid-level and senior developers, the requirements are higher. A mid-level should already be able to build their own workflow with AI, verify plausible but incorrect answers, and catch errors in complex business logic, large codebases, production incidents, and security issues. A senior is responsible for the next level: how to embed agents and code generation into the team process without increasing technical debt. Here, rules for using tools, context configuration for the project, reviewing AI-generated code, and clear distribution of responsibility are important, because the phrase "the model suggested it" doesn't work in production.

What this means

X5 Tech's approach shows a shift in hiring: the market increasingly tests less the ability to work "in a vacuum" and increasingly evaluates how an engineer acts alongside AI. Strong specialists are accelerated by such tools, while weak ones start to show gaps in understanding, architecture, and responsibility faster. For candidates, the conclusion is direct: it's important not just to use models, but to be able to explain your process, limitations, and decisions that you keep for yourself.