Workshop Programme

The workshops are fully booked and registration is no longer possible. Did you register for a workshop? You will receive your final ticket before the event!

Aurai

Data Engineering with Aurai

Participants will work with BAG Data and will work within a custom data platform solution that is internally developed. The participants need to answer several research questions with the help of the data. Hereby my colleagues will be presenting two real-life use cases showing how this concept is developed and used for external companies.

You need: a laptop and a email address.

Q42

Improving People’s Lives with AI: From Concept to Reality

Large language models have revolutionized the AI landscape, making it more accessible than ever to positively impact people’s lives. But where does one begin? In this workshop, you’ll learn how to turn your idea into reality.

You need: a laptop

This workshop is fully booked and only registered participants will be allowed. Note that we will start strictly on time and no late arrivals can be admitted.

CWI

Collecting the right data. A machine learning theory perspective on A/B testing

This workshop is aimed at prospective PhD students in machine learning. In it, we'll get a glimpse of the research process in action. We will start from an autonomous driving application, and ask where we should collect more training data to improve the system. We then distill this problem down to a more tractable abstract question, landing in the intersection of theoretical AI, machine learning, statistics and game theory. We'll then look at a possible solution, and reflect on what we learned.



We invite the audience to think along, and will answer questions about the content of the example and about the PhD process.

Institute for Logic, Language and Computation (ILLC)

Mechanistic Interpretability of Large Language Models: Opening the Blackbox

Transformers have emerged as dominant type of model architecture in AI. The inner workings of trained Transformers, like many other neural networks, remain opaque. Model-agnostic interpretability techniques are a popular approach to address this challenge, but their shortcomings are becoming increasingly clear. In this workshop, we  focus on Transformer-specific interpretability methods, including mechanistic interpretability techniques, that make use of specific features of the Transformer architecture. We focus on methods that combine causal methods with low-level Transformer analysis to identify particular subnetworks within a model that are responsible for specific tasks.

You need: There is a set of Jupyter notebooks and other materials ready at this site. Participants can thus immediately start working with the introduced techniques; however, the workshop can be followed without access to a laptop.

Databricks

DBRX: The model your data has been waiting for

Databricks will present the Data Intelligence Platform- the AI-driven evolution of the Lakehouse data architecture. Originally coined by Databricks, the Lakehouse paradigm is now endorsed by most enterprises and vendors globally. This enables organisations to implement all data driven use cases from streaming and batch ingestion, to data consumption for AI, BI, etc. The Data Intelligence Platform extends the Lakehouse by understanding your data, your workloads, coding style etc thereby greatly improving productivity. Also, it enables organisations to use or train their own models with their own data. This is exemplified by the development of DBRX on Databricks- constituting the current state of the art of open source LLM's.

You need: Your own happy self :)

Amsterdam UMC

AI from bench to bedside 

AI is advancing further and further and new applications are being developed quicker and quicker, with widespread implications. Prof. dr. Martijn Schut is keen on making AI models that are more understandable, but also on putting them to good use for clinical implementation. As a professor of Translational AI in Laboratory Medicine at the Amsterdam UMC he is the perfect candidate to tell you how to get healthcare AI from theory to practice.

BBO.life

'Bringing AI to Life

In this workshop BBO.life will elaborate on initial ‘best practices’ on what it takes to deliver value, for customers and other stakeholders, either hard (e.g. revenue, profit or GDP) or soft (e.g. NPS). They will explore some recent cases from Banking and Healthcare, where they will also explicate some differences in approaches, as well as some shared challenges. 

orq.ai

Productizing Generative AI from Playground to Production

In this workshop, you'll learn how to navigate Generative AI, taking your initial idea and guiding it through the steps to become a functional model in production. We'll cover the basics of Generative AI, breaking down complex concepts into manageable parts. Through practical exercises and guidance, you'll gain the skills needed to develop your use case from scratch. By the end, you'll have the confidence and knowledge to turn your ideas into tangible, deployable solutions using Generative AI techniques. Whether you're new to AI or looking to deepen your understanding, this workshop provides a structured approach to building your first Generative AI project.

You need: A laptop

ICAI

Solving inequality through AI: the stakeholder engagement PhD experience

In our session we present some of the work done at the Civic AI Lab, where our mission is to reduce inequality and increase equal opportunity through the proactive application of AI. The Civic AI Lab occupies a unique place in the ecosystem through its involvement with public sector actors like Gemeente Amsterdam and the Ministry of Interior. This comes with unique challenges like connecting theoretical research to real world applications, our PhD candidate Martin Smit will discuss what the study of multi-agent systems and reputations might tells us about inequality and its connection to institutional bias.