От LLM-основ до мультиагентных систем.
Последовательность из 12 курсов, сделанных как трек: каждый следующий опирается на предыдущий. Markdown-контент, практические задачи, и минимум воды — только то, что действительно нужно для работы.
Начать путь
Первые три курса — обязательные основы. Их же мы рекомендуем тем, кто приходит без LLM-опыта.
Introduction to AI Agents
Discover the world of AI agents — autonomous software systems that perceive their environment, reason about goals, and take actions to accomplish tasks. This beginner-friendly course covers what agents are, how they differ from standard chatbots, the main architectural patterns used in modern AI agents, and how to build your first working agent using LangChain. By the end you will have a solid conceptual foundation and hands-on experience constructing a simple but functional agent.
- ai-agents
- langchain
- beginner
- python
LLM Fundamentals
Build a solid understanding of how large language models work under the hood. This course demystifies transformers, tokenization, and attention mechanisms without requiring a PhD in mathematics. You will then apply that understanding to practical prompt engineering and learn to integrate LLMs via API — including OpenAI and Anthropic — with proper error handling, rate limiting, and cost management.
- llm
- transformers
- api
- openai
Prompt Engineering 101
Master the craft of writing prompts that reliably produce the outputs you need. This course covers core design principles — clarity, specificity, and role framing — before moving into advanced techniques like structured output coercion, constitutional AI constraints, and prompt chaining. The final module teaches you how to systematically evaluate and improve your prompts using evals and automated testing frameworks.
- prompt-engineering
- llm
- evaluation
- beginner
Как устроены курсы Kumo
Модули
Каждый курс разбит на 3–5 модулей. Модуль — законченная тема на 20–40 минут: прочитать + проверить себя на примерах.
Живой Markdown
Уроки — полноценные Markdown-статьи с кодом и диаграммами. Никакого видео без необходимости, всё фокусно по делу.
Локальный прогресс
Прогресс хранится в браузере — никаких аккаунтов, никакой привязки. Можно спокойно закрыть и вернуться.
Следующий уровень
Когда base-трек пройден — глубокие курсы про память, инструменты и мультиагентные схемы.
Tools and Memory for Agents
Learn how to give AI agents superpowers through tools and memory systems. This course covers function calling, custom tool creation, and the major memory architectures — short-term conversation buffers, vector-store long-term memory, and external structured memory. By the end you will be able to build agents that remember user preferences, retrieve relevant context from large knowledge bases, and use external APIs seamlessly.
- tools
- memory
- function-calling
- vector-store
Building Your First Agent
A hands-on project course where you build a complete, deployable AI agent from scratch. You will set up the project architecture, implement the core reasoning loop, integrate tools and memory, write tests, and deploy the agent. By the end you will have a portfolio-ready project demonstrating real agent engineering skills — not just theory.
- project
- deployment
- testing
- python
Advanced Agent Patterns
Go beyond basic ReAct agents to explore sophisticated architectures used in production AI systems. This course covers Plan-and-Execute, Reflexion, LATS (Language Agent Tree Search), and custom orchestration patterns. You will learn how to rigorously evaluate agent performance using benchmark datasets and implement production hardening techniques including circuit breakers, fallback chains, observability instrumentation, and cost controls.
- advanced
- plan-execute
- reflexion
- lats
Открой каталог полностью
12 курсов со всеми модулями и уроками — в платформе. Фильтры по уровню и тегам, прогресс между устройствами.