Applied LLMs & Agents

By the end of this winter school, participants will be able to:

• Set Up Modern AI Development Environments: Configure IDEs, virtual environments, and adopt "vibe coding" workflows to accelerate AI-assisted development.
• Understand Core NLP & Embedding Concepts: Grasp the foundational ideas behind classical NLP tasks, word embeddings, and vector search methods like cosine similarity.
• Navigate the LLM Ecosystem: Understand how Large Language Models work under the hood, their current capabilities, and how to identify high-value use cases for LLM applications.
• Build Custom RAG Pipelines: Design and implement Retrieval-Augmented Generation systems to ground LLMs in external knowledge, including bringing and processing your own datasets.
• Develop Agentic Workflows: Move beyond simple prompts to create AI agents capable of utilizing tools, skills, and advanced reasoning methods to solve complex tasks.
• Implement Systems Integration with MCP: Understand and build your own Model Context Protocol (MCP) server to seamlessly connect LLMs with external data sources and tools.
• Apply Fine-Tuning & Security Best Practices: Learn the basics of when and how to fine-tune your own models, while identifying and mitigating security vulnerabilities unique to Generative AI.
• Deploy Interactive AI Front-Ends: Rapidly prototype and share AI applications using Python-based UI frameworks like Gradio and Streamlit.

Monday (Arrival) 
14:00 - 14:30 Welcome and Check-In 
15:00 - 16:00 Set up environment (VENV, IDE, Vibe coding setup)
16:00 - 17:00 Introduction to NLP tasks
17:30 - 18:30 Apéro 
19:00 - 20:00 Dinner

Tuesday 
08:15 - 09:15 Word embedding – base idea, classical approaches
09:00 - 10:15 Word embedding  - advanced, embedding search – cosine similarity
10:15 - 10:45 Break 
10:45 - 12:30 Introduction to LLMs
12:30 - 17:00 Lunchbag, skiing, work or whatever 
17:00 - 18:30 LLMs applications
19:00 - 20:00 Dinner

Wednesday 
08:15 - 09:00 Integrated LLMs workflows – Intro to RAG
09:00 - 10:15 Integrated LLMs workflows – agentic RAG & advanced methods I
10:15 - 10:45 Break 
10:45 - 12:30 Integrated LLMs workflows – agentic  RAG & advanced methods II
12:30 - 17:00 Lunchbag, skiing, work or whatever 
17:00 - 18:30 Tutorial – Bring your own data - RAG 
19:00 - 20:30 Dinner (Regina) 

Thursday 
08:15 - 09:15 Integrated LLMs systems – MCP
09:15 - 09:45 Integrated LLMs systems – Skills
09:45 - 10:15 Break 
10:15 - 11:00 Security and Gen AI
11:15 - 12:30 Fine-tune your own model
12:30 - 15:00 Lunchbag, skiing, work or whatever 
15:00 - 16:00 Tutorial Write your own MCP server 
16:45 - 18:30 Sledge ride from Almendhubel down to the hotel, leaving the reception at 16:45 
19:00 - 21:00 Cheese Fondue at restaurant Alpenruh (meet at the hotel reception Regina at 18:45) 
21:30 - xx:00 Tächibar 

Friday 
08:15 - 09:00 Present MCP tutorial groups 
09:00 - 09:45  Front-end (Gradio - Streamlit)
09:45 - 10:15 Break and checkout 
10:15 - 11:00  Future Trends, Multimodality & The AI Horizon
11:15 - 12:00 Project and discussion session
12:00 - 12:30 Wrap up 

12:30         End of school 

The schools take place in the wonderful historic Hotel Regina in the center of the alpine ski resort Mürren.

Mürren is situated on a mountain terrace 1638 meters above sea level in the Bernese Oberland, granting an unique view of the majestic mountain range of Eiger, Mönch and Jungfrau rising on the opposite side of the narrow mountain valley. Mürren is car free.

Accommodation costs at Hotel Regina Mürren include a room with shared bathroom, breakfast, coffee break, lunch bag, dinner, and social program for the duration of the Winter School. Additional consumption at the hotel needs to be paid directly upon check-out. 

Participants who participate remotely only pay the course fee. 

Payment

By submitting the registration form, you are making a binding registration for the Winter School.

After your registration has been received, an invoice for 50% of the total participation fee (course fee + accomodation) will be issued. The remaining 50% will be invoiced two weeks before the start of the Winter School. Payment must be made by the due date indicated on each invoice.

Cancellation Policy

Cancellations must be submitted in writing by email to info.dsl@unibe.ch.

• Until 15 February 2027: Cancellations are free of charge. Any payments already received will be refunded in full.
• After 15 February 2027: Cancellations are possible; however, refunds will be subject to a CHF 150 administrative fee.
• No refunds will be granted for participants who fail to attend the Winter School without prior written cancellation (no-shows).

Refunds

Refunds will be processed after the Winter School has concluded.

If the Winter School is cancelled by the organizers, all payments received will be refunded in full. The organizers accept no further liability beyond the reimbursement of the participation fee.