What we’re about
Regensburg is full of data science expertise, both in industry and in academia. Our aim is to bring together people who share an interest in this area and offer an environment for networking in an informal setting. We continue to have speakers with a range of backgrounds offering insights into a wide spectrum of data science ranging from enterprise search to music recommendation, from automatic fact-checking to avoiding harms and biases, from generative approaches to automatic question-answering. And that is not even everything. Other topics include large language models, industry use cases of natural language processing and the list goes on and on ... We have speakers from industry (e.g. Bloomberg, Netflix, Amazon, Spotify, Deloitte ...) and universities (CMU, Queen Mary, Essex, Regensburg ...). Want to present? Drop us a message. For more details on the organising team check: https://ai.ur.de/
Upcoming events (2)See all
- A Deep Dive into Domain-Adaptation for Generative AI ApplicationsUniversity of Regensburg, Regensburg, BY
What's going on? Just announced our roadshow that takes us to Berlin and here we run another event? Well, why not? You deserve it!
We managed to line up an excellent speaker on an exciting topic. We are looking forward to an evening of insights around all those recent developments that we cannot keep track of... LLMs, prompting, fine-tuning, generative AI, domain-specific (real-world) NLP applications (and the list goes on). Aris will tell us what really matters and he can do so as he knows what works and what does not work in industry. We are looking forward to welcoming you on campus next month.
Udo, David and Bernd
P.S.: The trip to Berlin in April is shaping up. We have a room, we have drinks, we have a speaker (details to be revealed soon, only this much: another real-world problem of data science to be addressed) ... and we have a cohort of around 10(!) data science enthusiasts travelling to Berlin from Regensburg ... that's what I call dedication.
Aris Tsakpinis (Specialist Solutions Architect for AI & Machine Learning with Amazon Web Services)
Align your LLM-powered application right - a deep dive into domain-adaptation for generative AI applications
Aligning a LLM-powered application to a use case domain is a crucial element of building helpful, harmless, honest and thus enterprise-grade applications. In this meetup's session Aris will dive deep into various fine-tuning variations like continued pre-training, supervised fine-tuning or reinforcement learning based approaches. He will also shed light into different approaches of dynamic prompt engineering for domain-adaptation, like retrieval-augmented generation or the concept of LLM-powered agents. Finally, he will provide guidance on which approach to choose best given specific application areas considering functional and non-functional requirements, based on state-of-the-art research findings and empirical observations from industry.
Aris Tsakpinis is a Specialist Solutions Architect for AI & Machine Learning with Amazon Web Services, focussing on natural language processing (NLP), large language models (LLMs) and generative AI. In addition, he is pursuing a PhD in ML Engineering at University of Regensburg, researching the field of applied NLP in the science domain.
- Data Science @ Regensburg goes Berlin: iSchool SpecialDepartment of Library and Information Science (Humboldt-Universität zu Berlin), Berlin
Need a reason to plan a trip to Berlin? The wait is over. It is a pleasure to announce our Data Science @ Regensburg Meetup Roadshow. Apart from the location the setup is as always: a relaxed forum to discuss data science issues with a broad mix of attendees from industry, academia and beyond. What's more, this event also represents the first activity we run as part of our European iSchool Industry agenda (https://www.ischools.org/industry-partnerships) .
And here is our first confirmed speaker: Alexander Buchholz (Senior Applied Scientist at Amazon). More details below ...
See you in Berlin?
Alexander Buchholz (Senior Applied Scientist at Amazon).
Amazon Music ranks millions of items to personalize its customer experience. In this talk we will provide a perspective on the scientific questions that we encounter for training and evaluating learning-to-rank models. We will touch upon topics like unbiased off-policy learning and evaluation, data collection and model deployment from an industry point of view. We will highlight our latest contributions and discuss the challenges that we face on a daily basis.
Alexander Buchholz is a senior applied scientist at Amazon Music in Berlin. Prior to joining Amazon Music he was a postdoc at the University of Cambridge, UK. He received his PhD in computational statistics from the University Paris Saclay and studied Economics, Mathematics and Statistics in Berlin, Paris and Cambridge, MA.