Ukraine’s first summer school for those who want to do more than just use AI; they want to build autonomous AI agents and systems that tackle real-world problems.
Why KSE Agentic AI Summer School?
9 weeks of intensive practice
Intensive work with autonomous agents on real-world problems.
GPU infrastructure included
Access to server resources and the agent runtime for running models, experiments and prototypes.
Free participation
You don’t simply buy your way into the programme. You go through a competitive selection process and demonstrate that you are ready to develop competitive solutions.
Scholarship
$600 per month for programme participants.
Real-world challenges
Case studies from partner organisations.
Demo Day
Final presentation of team solutions to technical teams, partners and potential users.
Programme format
Nine weeks of study and work on a team project.
Final Demo Day featuring project presentations
4 sprints with separate demo sessions and a review of the results
Regular office hours with lecturers from various disciplines
Weekly face-to-face sessions and project work
Additional modules for self-study (Machine Learning, Deep Learning, mathematical foundations)
Lectures and practical sessions, consultations, and reviews of homework and project assignments
9 weeks, 4 sprints
Sprint 1 (Weeks 1–3): From getting to know the LLM to your first agent
● Week 1: The LLM landscape, how modern LLMs work, an overview of platforms and tools
● Week 2: Prompt engineering, context, RAG, security, and prompt injection
● Week 3: Tool use, function calling, MCP servers, integration of external systems
● Sprint Demo #1
Sprint 2 (Weeks 4–5): Agent Architecture and Memory
● Week 4: Agent architecture and orchestration (ReAct, reflection, planning, multi-agent, LangGraph, PydanticAI), state management and human-in-the-loop
● Week 5: Reasoning, planning, memory, and long-term planning and memory in agents
● Sprint Demo #2
Sprint 3 (Weeks 6–7): Agents that write code and learn
● Week 6: Coding agents, software development using agents, and code verification
● Тиж. 7: Reinforcement learning, preference learning та self-improvement loops
● Sprint Demo #3
Sprint 4 (Weeks 8–9): The Road to Production
● Тиж. 8: Evaluation, safety, Responsible AI та deployment
● Week 9: Production hardening, observability, documentation and pitch preparation. Final Demo Day
This is where Ukraine’s first generation of AI architects is being trained
In the 1990s, KSE trained the first generation of economists who built the modern Ukrainian economy — the National Bank, reforms and market institutions. The same logic applies here.
The first generation are the people who bring about change in institutions. Not because there are so many of them, but because they set a new standard. We want to establish such a standard for leaders in AI.
What will you be after the programme?
AI Agent Builder
You will be able to create agents capable of independently performing complex multi-step tasks: collecting data, testing hypotheses, invoking tools and generating results.
AI Systems Architect
You will learn how to design the logic of autonomous systems: how an agent makes decisions, where human intervention is required in the system, how to minimise errors, and how to make the system suitable for real-world use.
AI-Native Problem Solver
You’ll be able to take a complex challenge in education, business, analytics, engineering or the public sector—and turn it into a working AI prototype.
Builder with Demo Day Portfolio
By the end of the course, you’ll have not only a certificate but also a final project that you can present to technical teams, partners and potential users.
By the end of the course, you will be able to
Formulate tasks for AI systems, rather than just writing prompts
Working with logs, errors and model limitations
Developing prototypes of AI tools for real-world tasks
Build agent workflows consisting of several steps
Check the quality of the result rather than relying on the first answer
Present the solution to users and technical teams
Stages of the competitive selection process
We don’t look at formal qualifications, but rather at how you think, how you work with AI tools, how you iterate, how you spot mistakes, and how you deliver useful results.
Stage 1
Request with an AI artefact
Please send us a transcript of a real conversation you’ve had with an AI tool—such as ChatGPT, Claude, Gemini or another—over the past month. It should be a moment when you did something useful for yourself or for others.
Please submit an essay of up to 250 words in English: describe what you did, where the model went wrong, how you adjusted your approach, and what you learnt from the experience.
Once you have submitted your application, please wait for the results of the first stage to be sent to your email.
Stage 2
Online session
If your application has passed the first stage, we will invite you to an online session. It consists of several parts:
A practical exercise using AI tools — a real-world task for which you will be given time and access to any AI tools. We are not looking for a perfect answer, but rather the thought process: how you iterate, correct mistakes and refine your approach.
The technical section consists of questions testing mathematical and logical reasoning, as well as a basic understanding of how algorithms work. We do not require any programming experience, but we do require the ability to analyse code.
Stage 3
One-to-one interview
A final short interview in English for candidates who have passed stage 2.
We’re not interested in a prepared answer, but in how you explain your own thought process at that specific moment and how you work with AI in real time.
Stage 4
Start of the course
The course takes place in person over the summer in Kyiv
What we assess
Thinking as a process
Whether you accept the initial AI response as final, or iterate and improve the result.
Handling errors
Whether you identify the model's weaknesses, verify facts, and adjust your approach.
Practical outcome
Whether you can create an artifact that is truly usable.
Autonomy
Whether you can independently organize the AI workflow and drive the task to completion.
Certificate following the defence of the final project
Participants who complete the programme and successfully present their final team project at Demo Day will receive a KSE Agentic AI Summer School certificate.
The certificate confirms completion of an intensive practical programme in AI agent development, working on real-world tasks, and the presentation of the final solution.
Frequently Asked Questions
What sort of scholarship is it?
A scholarship of $600 per month
What criteria and skills are prioritised for admission to the programme?
To be accepted onto the programme, it is not formal qualifications that matter most, but the ability to think, learn and work with AI as a tool. We look for candidates who can tackle a problem independently, iterate and improve results, spot errors in the model, verify facts, ask more precise questions, and turn their work into a practical artefact that can actually be used. Also important are basic mathematical and logical thinking, the ability to reason about algorithms and code even without extensive programming experience, as well as a sufficient level of English to work with materials, participate in interviews and explain one’s own thought process.
In which language are the classes taught?
English and Ukrainian
Can you take the course if you don’t live in Kyiv? How many sessions do you need to attend?
Students from outside Kyiv are provided with accommodation assistance for the duration of their studies at the school. In-person attendance is compulsory.
Do I need paid AI subscriptions whilst studying?
Students do not need a subscription. Students are granted access to GPU resources and API subscriptions.
How will the final certification and project presentations take place?
There are no exams in the programme. Assessment is based on sprint demos, work on the team project, contribution to the team’s work, and a final presentation of the results. At the end of the programme, teams present their project at Demo Day: they demonstrate the finished solution, explain key design decisions, show how they tested its quality, and discuss limitations and opportunities for further development.
Who teaches on the programme?
The sessions are led by KSE lecturers and international speakers from leading companies with relevant practical and research experience in developing AI systems, working with modern LLM tools, and implementing technological solutions to real-world problems. Individual sessions are also led by industry leaders, practitioners from companies, and guest speakers from partner organisations.
The programme will be useful for:
• Year 11 pupils and school leavers with a strong aptitude for mathematics;
• students of mathematics, computer science, physics and engineering;
• students of economics, business and public policy who wish to gain practical experience working with AI systems;
• engineers, analysts, physicists and other specialists who are used to solving complex problems and are ready to work on real-world projects.