“AI training has already become mandatory for all specializations and levels, including beginners”: how EPAM, Ciklum, SoftServe and Sigma Software are training specialists in the fall of 2026

Before moving on to the programs themselves, it is worth making a reservation right away: not all of them are open to outsiders. At Ciklum, for example, there is currently no external training at all — the company directly says that all AI Learning Programs are designed exclusively for current employees. And this is not the only case of such an approach, although it is the most consistent. At other service companies, external and internal training coexist, albeit in different proportions. For example, EPAM maintains a separate public portal for candidates campus.epam.ua, and Sigma Software University was built from the very beginning for both external and internal audiences — more than 100 programs and about 5,000 participants every year.

Who trains newcomers?

EPAM works most systematically with candidates «from the street». In a year and a half, Iryna Sokolova, head of the training and development department at EPAM Ukraine, told dev.ua, over 5,000 participants have gone through the multi-level EPAM Campus programs — Basic, Beginner, Fundamentals, Specialization — and about a thousand of them have reached the final stage of specialization. As a result, last year over 600 juniors joined the company, this year — over 200. The logic of this approach is explained by the company simply: it is more effective to train a strong university graduate than to look for a ready-made junior on the market, who will still have to be trained and adapted.

Sigma Software implements the same idea — only under the name of «internships.» Last year, the company reviewed over 300 applications, 58 interns received offers to join the staff, and this year there are already 23 new ones.

A parallel mechanism at SoftServe is called SoftServe Academy: in 2025, 580 graduates graduated from hiring-track programs, 260 of whom became part of the team.

Thus, the three companies, despite their different names and sizes, essentially solve the same problem — to prepare newcomers in advance for their own needs, rather than competing for ready-made Junes on the labor market.

Development, QA, DevOps and Cybersecurity

When it comes to purely technical tracks for current or future engineers, SoftServe offers the most comprehensive set among the companies surveyed: separate programs in Python, Java, JavaScript, .NET, DevOps, Cybersecurity, Test Automation, QA, and Frontend/WebUI. But these are not the same courses as they were a year or two ago — the company emphasizes that «the modern developer/tester/devops actively manages agents in their work to solve business problems,» so AI is woven into even classic technical tracks, rather than being relegated to a separate course.

A similar movement at Sigma Software is visible not through the list of technologies, but through specific career transitions that the company supports: a Software Engineer gradually becomes an AI Engineer, a QA Engineer moves to Automation QA or AI QA, a Business Analyst to Data/AI Analyst, and a Data Engineer to ML Engineer. Under these trajectories, Sigma Software University has separate programs in Data Science, Big Data, BI & Data Analytics, AI Engineering, and LLM Engineering — that is, where SoftServe updates existing tracks from within, Sigma Software rather builds separate parallel routes for the same transitions.

Management and leadership

Ciklum, despite the lack of open training for an external audience, has an internal management ecosystem. The flagship GEM leadership development program of the last cycle selected 30 high-potential leaders, another 120 managers completed programs for New and Middle Managers from various delivery centers, 54 Delivery Managers participated in Delivery Capability development programs, and 39 specialists in Product Management and Business Analysis participated in the Strategic Communication & Influence Program.

At the same time, the company is quite cautious in formulating the connection between training and career: «we are cautious about directly attributing the success of a career transition solely to training, since a whole range of factors influence professional development» — although they admit that within the framework of GEM, 10 program participants did receive promotions.

Artificial Intelligence Courses: What You Can Start in the Fall

If you are looking for something with specific, already known start dates among all the areas, then the easiest way to find your way around is Sigma Software: AI-Native Engineering starts on October 16, and AI-Assisted Engineering starts on October 21.

SoftServe does not have a separate «AI course» in fact — instead, there is a whole architecture of three parts. The first is about building AI systems: AI Development Fundamentals, Advanced Python for Data Science and Big Data, Advanced Data Science and Machine Learning, Modern MLOps/LLMOps/AgenticOps and project-based learning, combined into a cross-curricular program «Complete course AI & Data Science Engineer». The second is the already mentioned updated classic tracks. The third is purely applied: the course «Artificial Intelligence for working with documents» on how to delegate routine work with documents, tables and reports to AI.

According to Halyna Datsiv, Vice President of Training and Human Resources Development at SoftServe, the company has changed not just the program, but the training model itself — «from transferring knowledge about AI to developing the ability to solve specific engineering and business problems with its help.»

The company says that they have already passed the level of building general awareness: about 80% of the staff (7922 people) are involved in the development of AI competencies, 77% of them have completed at least one course, and in the Delivery direction, training has covered more than 90% of the organization. The ecosystem has more than 20 practical use cases, 30 specialized GenAI programs and more than 130 AI experts. This has even resulted in the emergence of new positions: Intelligence Engineer, who configures and develops AI agents and integrates them into development processes, and Intelligence Architect, responsible for the architecture and scaling of AI solutions. According to the company’s SDO director Ivan Zagorodny, over the past year more than 250 employees have acquired this new specialization.

Against this background, EPAM’s approach looks almost mirror-like: the company fundamentally does not offer a separate external course on AI, instead, the AI ​​component is already built into every specialization on campus.epam.ua. According to Iryna Sokolova, the company has instead launched a large-scale internal program for its own specialists, AI/RUN Mission 2026, focused on building an «AI Factory» — an end-to-end, AI-native delivery process that is still managed by people. In Ukraine, more than 1,800 colleagues, or approximately 21% of the company’s technical specialists, have already joined it.

At Ciklum, all AI content is purely internal, but built on a similar logic of gradual immersion: the three-level Ciklum AI Learning and Certification Program takes employees from a basic understanding of AI, LLM and AI tools through practical application in daily work to in-depth work with AI agents. At the second level, the program branches into separate learning paths for Developers, DevOps, QA, Business Analysts, Delivery and other roles — that is, AI is not a separate profession here, but a layer of skills that is superimposed on the already existing human expertise. During this period, more than 2,500 Ciklum specialists have undergone AI training at various levels — from basic familiarity with technologies to practical application.

Global AI Course

Despite different formats and different openness to an external audience, all the companies surveyed describe essentially the same shift: AI training has ceased to be a separate optional course and has become a mandatory part of training — for both beginners and existing specialists. Iryna Sokolova from EPAM puts it bluntly: «AI training has already become mandatory for all specializations and levels, including beginners.» And Sigma Software draws attention to the other side of the same process — the speed with which knowledge becomes obsolete: «what was relevant to study a month ago is already outdated today,» so the company relies not on long manuals, but on short and fresh practical lectures for specific tools. Together, these two remarks, in fact, describe the same reality from two different sides — the mandatory nature of training compensates for its rapid obsolescence.

Remote, hybrid or office? How Ukrainian IT companies choose the format of work in conditions of alarms and shelling

On the topic

Remote, hybrid or office? How Ukrainian IT companies choose the format of work in conditions of alarms and shelling

Read the country’s main IT news in our Telegram

On the topic

Read the country’s main IT news in our Telegram