RAG, Generative AI, and Fine Tuning - The OpenSearch Meetup Amsterdam
Zeta Alpha invites the Gen AI and Search community in Amsterdam to join our next meetup!
We'll kick off at 18:00 with Pizza, Refreshments, Talks, and Networking (this is an in-person only event).
This time we have a packed lineup in a more central location, featuring speakers from AWS, Eliatra and Zeta Alpha. Don't miss out!
What our group is about:
We are a community in the Amsterdam area with keen interest in Generative AI, Search Engineering, and Open-Source Technology. Whether you are building production systems or simply curious about the latest and greatest in AI and Search technologies -- such as Agents, RAG, Fine-Tuning, Vector Search -- this is the space for you!
We welcome all skill levels. Come learn and connect with other smart engineers.
Time and location:
📅 November 13th, 2024 from 18:00 to 20:00.
📍AWS Office, Mr.Treublaan 7, 4th floor, 1097 DP, Amsterdam.
Schedule:
18:00 -- Arrival. Grab some 🍕 and 🥤.
18:15 -- Maximizing Your Hardware Potential With Local LLMs on OpenSearch - Lucas Jeanniot (Eliatra)
18:35 -- TBD - Cédric Pelvet (AWS)
18:55 -- Build and scale your generative AI application the right way - Maurits de Goot (AWS).
19:15 -- Creating state-of-the-art embedding models - Arthur Câmara (Zeta Alpha).
19:35 -- Networking with drinks.
20:00 -- Closing.
Talks:
Maximizing Your Hardware Potential With Local LLMs on OpenSearch
In this talk, we will explore how to maximize your hardware potential by integrating local Large Language Models (LLMs) with OpenSearch, creating efficient and scalable AI-driven search experiences. We’ll start with an overview of Retrieval-Augmented Generation (RAG) and how it enhances the power of traditional search by using AI to provide more relevant and context-aware results. From there, we’ll dive into implementing local LLMs within the OpenSearch ecosystem, focusing on practical steps for connecting external models using Connector Blueprints. Attendees will also learn how to set up and optimize search and ingest pipelines, leveraging the full potential of their hardware resources for AI-powered search solutions. This session is designed to be hands-on and approachable, making it ideal for AI enthusiasts looking to deepen their OpenSearch knowledge and push the boundaries of what’s possible with local LLMs.
Speaker: Lucas Jeanniot, Machine Learning Engineer at Eliatra.
Tips & Tricks for Vector Workload Optimization
OpenSearch makes it easy to get started using vector search, but using it efficiently is a different story. Let’s talk about the key optimization techniques to get the most out of OpenSearch.
Speaker: Cédric Pelvet, Principal Specialist Solutions Architect at AWS.
Build and scale your generative AI application the right way
Building a generative AI application can be tricky. Besides making your awesome application you want to ensure that your data is secure, your application scales well and that you have the right performance. In this 20 minute session you’ll learn more about the different deployment strategies, and implications for your GenAI application.
Speaker: Maurits de Goot, Solutions Architect at AWS.
Creating State-of-the-Art Embedding Models
While encoder-only models have dominated the embedding landscape, a new generation of decoder-only embedding models, based on pre-trained LLMs, is quickly emerging as a powerful alternative to BERT-like models. In this talk, we will talk about how to fine-tune an embedding model and how to build a model based on existing LLM models. We will go through the process of the creation of the Zeta-Alpha-E5-Mistral, Zeta Alpha’s first open embedding model, and how we plan on continue pushing this direction with more efficient and performing models
Speaker: Arthur Câmara, Research Engineer at Zeta Alpha.