Kazakhstan is facing a critical shortage of professionals who can make AI work for real business and government needs.
The demand for AI engineers is growing faster than specialists can be trained. In this article El.kz explains how to enter a profession that already pays millions, and why there is such a shortage in Kazakhstan.
2026 has been officially declared the year of digitalization and AI development. Companies and government agencies are actively implementing neural networks from analyzing bank transactions to predicting retail demand. However, the number of people capable of configuring and maintaining these systems is low.
The shortage of middle- and senior-level specialists is particularly acute. Almost half of Kazakhstani companies cite a lack of qualified AI personnel as the main barrier to technology adoption. The state is investing in digitalization, but without engineers capable of turning pilot projects into working systems, everything risks remaining just "pretty presentations."
An engineer trains neural networks on a company’s specific data, builds systems that can answer customer questions based on internal documents, and optimizes computing costs. Yesterday, they might have been setting up a chatbot for a bank. Today implementing product recommendations for an online store. Tomorrow helping doctors identify pathologies on medical scans faster.
A junior AI engineer in Kazakhstan can expect 600,000 tenge per month. A specialist with three to five years of experience can easily earn 1–1.5 mln. Senior and lead engineers at top companies or in remote roles earn between 2 and 2.5 mln tenge and above. For comparison, this is several times higher than the national average salary.
Globally, the figures are even higher. In Almaty and Astana, experienced specialists are already approaching pay levels that were previously considered reserved only for expats. At the same time, competition remains fierce: there are often more than ten applicants for every high-quality job opening.
Today, companies pay not for the number of frameworks you’ve memorized, but for the ability to deliver a working product. Python remains the primary language. PyTorch or TensorFlow are used for deep learning. It is crucial to understand vector databases and know how to build RAG (Retrieval-Augmented Generation) systems, where the model doesn't just answer "off the top of its head" but relies on the company’s actual documents.
Another in-demand skill is MLOps, the ability to deploy models into production, monitor their performance, and quickly fix issues if something goes wrong. Without this, even the smartest model quickly turns into an expensive toy.
But there is a downside. Technology changes so fast that yesterday’s knowledge can already be outdated. You have to learn constantly. Many burn out because, after work, they still need to follow new research papers and experiment with new tools.
Competition is also a factor. Remote vacancies from other countries sometimes offer more comfortable conditions. Therefore, Kazakhstani companies have to fight for talent not only with money but also with interesting challenges.
If you are ready for this pace and want to be the person who turns "smart future" ideas into real systems, the path is open. The main thing is to start not with loud promises, but with daily practice and an honest assessment of your skills.
While the market is not yet saturated, there is a window of opportunity. In a couple of years, entry will be more difficult. But the value of those who enter now and gain real experience will only continue to grow.