Company Presentation: "LLM-as-a-Judge" with UBS

"LLM-as-a-Judge" refers to the use of Large Language Models to evaluate and judge the quality of generated content or responses in various applications, including the performance of other AI models. In this hands-on, interactive event, we will explore the practical applications of LLMs. We will demonstrate how to use open-source software to download and run models, integrate them into chatbots, and analyze outputs.

Topics covered will include prompt engineering, addressing model inaccuracies, and implementing Retrieval-Augmented Generation (RAG) for improved accuracy. We will also demonstrate the use of LLM-as-a-Judge in Chain-of-Thought (CoT) and agent-based solutions, while touching on security and guardrails.

All materials and code are publicly available, showcasing the current capabilities of LLMs.

With the participation of:

• Flavia Donno, AI Enablement Chapter Lead, UBS
• Mihai Gurban, Senior Data Scientist, UBS
• Matthias Ramirez, Data Scientist, UBS
• Andreia David, Business Analyst, UBS (USI Alumna, Bachelor in Informatics, 2009, Master in Software Design, 2012)
• Davide Mammarella, Software Engineer, UBS (USI Alumnus, Master of Science in Software & Data Engineering, 2023)

Introduced by:

• Ernst Wit, Dean Faculty of Informatics, USI

Date: 6 October 2025, H. 12.40-13.20, A brown bag will be provided to participants
Location: Room C1.03, East Campus, Lugano
Language: English

To participate:

• Registration is closed. If you would like to participate, we kindly invite you to contact [email protected]