Munich MLOps Community Meetup #6

Hello, fellow MLOps Engineers and ML Enthusiasts, and happy new year!

We're announcing our next meetup event - we're back for our first one in 2024 on Tuesday, January 30th at 18:30, at adesso office spaces (close to Leuchtenbergring S-Bahn station).
This time we'll hear from Saumya and Stefan about their experiences with open-source tools and models - from optimizing advanced Machine Learning pipelines to LLM fine-tuning with concrete examples based on NVIDIA HGX™ H100 and Intel Gaudi2 accelerators!

Full agenda announced:

18:30 - 18:50 - Beers & Networking
18:50 - 19:00 - Opening talk from the host of the event - Sashel Niles Gruber, Competence Center Manager at adesso
19:00 - 19:40 - Talk #1: Leveraging Open Source Tools for Advanced Machine Learning Pipelines - Saumya Goyal, MLOps Engineer extending consultancy to BSH
19:40 - 20:20 - Talk #2: Unleashing Humor with AI: MemeGPT - Fine-Tuning LLMs for Meme Generation using NVIDIA HGX™ H100 and Intel Gaudi2 - Stefan Ojanen, Director of Product Management at Genesis Cloud
20:20 - Onwards - Beer, food & networking

More about the talks:

• Speaker #1:
  Saumya Goyal, MLOps Engineer extending consultancy to BSH (LinkedIn)
• Title:
  Leveraging Open Source Tools for Advanced Machine Learning Pipelines
• Description:
  Immerse yourself in a dynamic presentation centered around optimizing the Machine Learning project pipeline through the strategic utilization of open-source tools. Our discussion will traverse the landscape of seamlessly integrating diverse tools to construct an efficient and streamlined machine learning development pipeline.
• Key Points:
  Open Source Mastery: Explore the pivotal role of open-source tools in revolutionizing the Machine Learning ecosystem, unlocking new dimensions of efficiency and innovation.
  Tool Showcase: Delve into the specifics of LabelStudio, a powerhouse for labeling, and Voxel51, a key player in Explainable AI. Uncover their functionalities and understand how they contribute to elevating your ML projects.
  Enterprise Deployment: Gain insights into the deployment intricacies of these tools at an enterprise level on cloud platforms. Witness firsthand how scalability and performance are optimized in a cloud-centric environment.
  Crafting Cloud-Agnostic Pipelines: Learn the technical art of constructing cloud-agnostic pipelines, offering flexibility and freedom from vendor lock-ins. Explore strategies to ensure your ML pipeline remains adaptable across various cloud platforms.
• Speaker #2:
  Stefan Ojanen, Director of Product Management at Genesis Cloud (LinkedIn)
• Title:
  Unleashing Humor with AI: MemeGPT - Fine-Tuning LLMs for Meme Generation using NVIDIA HGX™ H100 and Intel Gaudi2
• Description:
  Join Stefan in a captivating journey through the realm of AI-driven meme generation. In this talk, we explore the intricacies of developing a custom Large Language Model specifically for creating humoristic memes. Gain firsthand knowledge on fine-tuning an open-source LLM using cutting-edge NVIDIA HGX™ H100 and Intel Gaudi2 accelerators. This exploration includes understanding Parameter-Efficient Fine-Tuning (PEFT) methods like LoRa and QLoRa, integral to enhancing the model's capabilities. We'll delve into the linguistic challenges and the humor elements that make memes so appealing and relatable. The session offers an in-depth look at how advanced technologies, linguistic understanding, and innovative methods converge to create engaging, humorous content.
• Key Points:
  Build and Fine-Tune LLMs: Learn to construct and enhance LLMs using NVIDIA HGX™ H100, Intel Gaudi2, and PEFT methods like LoRa and QLoRa.
  Linguistics and Humor in AI: Explore the challenges of integrating humor in AI-generated content, focusing on memes.
  AI-Generated Meme Showcase: Experience practical examples of small and large-scale Language models in meme generation.
  Recent AI Developments: Understand the latest advancements in AI content generation and the importance of DIY approaches

Looking forward to your RSVPs and to meeting you there!

Keep on hacking!🤩🤩🤩