
Recent advances in AI, particularly the emergence of agentic systems capable of autonomously executing complex tasks, are transforming scientific research by making computational analyses more accessible and enabling new ways of conducting research.
This course offers an introduction to AI-assisted research, focusing on the practical use of AI agents within the university's computing infrastructure. Practical demonstrations using Claude Science, an AI agent for scientific research, will show how to access the tool, conduct scientific analyses using university computing resources, and connect AI agents to external tools, databases, and services. A sandboxed variant of Claude Science deployed at the university’s infrastructure will be described as a tool mitigating data safety and security issues related to agentic systems.
The course is in English and will be streamed online.
Registration is open on this link.

The goal of this hands-on course is to show researchers how to extend AI agents with custom tools and knowledge, and how to connect them to their own research software and workflows.
This is essential for turning LLMs into practical tools for science and education. We will explain and practise working with concepts such as the Model Context Protocol (MCP), Agent Skills, and retrieval-augmented generation (RAG).
Target group: Pedagogical staff, PhD students, DPP Employee
Language of instruction: English
Lecturer: Ing. Sára Veselá
Venue: Lecture Hall T8, Faculty of Mathematics and Physics, Charles University
Registration is open on this link.

The MIRACLE Winter Academy is a four-day course in data analysis and AI-assisted research for PhD students working with biomedical data.
Unfortunately, the course is fully booked.

Profiling and optimisation in Python". Anotace: "Participants will learn how to identify bottlenecks in Python code using profiling tools, and how to speed it up with vectorisation, compilation (Numba, Cython) and multi-threading.
Selected pitfalls that beginners often run into will be demonstrated, together with ways to improve code performance. By the end of the course, participants will be able to identify bottlenecks in their code and optimise it.
Language of instruction: English
Lecturer: Mgr. Martin Topinka, Ph.D.
Registration is open on this link.

Learn how to use AI coding agents to build scripts, tools, and small apps for your research, even without a strong programming background.
In this 3-hour hands-on workshop, you’ll practise a simple workflow: plan first, give the model the right context, and build in small, tested steps. We’ll work with opencode connected to Charles University’s LLM gateway.
The focus is on a simple way of working: plan before you build, give the model the right context, and grow your project in small, tested steps instead of piling on features. Beyond the workflow, we will look at what the coding agent itself offers and how to extend it: slash commands, project instruction files (AGENTS.md), skills, and MCP servers.
We will work with opencode, an open-source coding agent, connected to Charles University’s LLM gateway (LiteLLM). All participants will receive access and setup instructions before the workshop. You may use your own tool (e.g. Claude Code, Cursor, GitHub Copilot), but the demonstrations and technical support will be based on opencode.
Language of instruction: English
Lecturer: Luděk Svoboda
Registration is open on this link.