Finam Engineer Maps the Russian LLM Development Market: From RAG to Salaries

A senior software engineer at Finam and author of the "LLM Developer" course at Hexlet has published a practical overview of the LLM development market in Russia on Habr. His timeline goes like this: the world more or less understood where AI was heading around 2024, while Russian business "came to its senses" in 2025–2026. The demand for adoption is there. Often, what exactly to adopt is not.

Demand Outpaces Task Definition

The author describes a scene many engineers will recognize. The CEO comes to a developer and says: we need AI everywhere. The company wants to build it into several processes at once:

• development;
• CI/CD;
• task estimation;
• code review;
• task definition;
• work with assistants.

According to the author, AI really can be applied in each of these areas, even to the point that a CEO could order his coffee through a chatbot. The problem lies elsewhere: the business cannot say clearly which metric the AI is supposed to improve. The revolution, in his words, is real, but every revolution has to be steered: you need to understand how, why, and where to apply it.

Tellingly, the author wryly addresses the terminology: AI, he says, is "not quite intelligence," but nobody cares, and everyone wants to use it. For an engineer, this means a significant part of the work begins before any code is written: you have to turn a vague "we want AI" into concrete scenarios with measurable results.

Which Questions the Author Sets Out to Cover

The introduction lists a set of practical topics. Who actually needs RAG. Whether you have to deal with large models at all. What is happening in the LLM development market. How to get into the profession: which skills are needed, how to look for a job, and how much to ask for in an interview.

The available excerpt does not answer these questions: the text cuts off at a "Read more" button. It contains no salary ranges, no list of technologies, and no analysis of specific cases. Judging by the teaser, this material is in the full version of the article and in a video format that the author posted separately.

How to Read the Piece

The text was published by the author of the course and includes a link to the training program with advertising markup. That does not negate its practical value: a person working as a senior engineer at a fintech company sees real business requests. But such material has an obvious interest behind it, and market figures are best checked against independent sources: job postings, recruiter reports, and developer surveys.

The practical takeaway for different readers is simple. An engineer who wants to move into LLM development should start by asking what problem the system solves and how its quality is measured. The model and the framework are secondary. A manager preparing an order to "adopt AI everywhere" would do well to pick one or two processes with a clear metric in advance, such as code review time or the share of automatically resolved requests, and start there. Those considering a career change should read the full text and check the stated requirements and pay expectations against live job postings before building a study plan.