MLOps Community San Francisco Fall & Winter Workshops - Part II
Are you building a GenAI-powered Product? If you've been tinkering with an idea, created a prompt-based agent prototype, a simple Retrieval Augmented Generation (RAG) agent, or even fine-tuning your own LLM in Jupyter Notebooks, and want to learn best practices in taking your prototype to production, this workshop is for you!
In part II of our "Fall & Winter Workshops" series, the MLOps.community is partnering with Verta to cover some best practices and skills you'd need for taking GenAI from prototype to production. These workshops are for Software Engineers, AI/ML Engineers, and Product Managers trying to figure out how they productize the GenAI prototypes they've built.
This will be a half-day event with two sessions:
1:00pm - 3:00pm: Your LLM Stack from Prototype to Production - design considerations and best practices for production-grade LLMOps for RAG applications, fine-tuning models, and more by MLOps Community.
3:00pm - 3:30pm: Break
3:30pm - 5:30pm: GenAI Application Prototype to Production: From tinkering with GPT to Shipping GenAI apps with high-quality and low-cost by Verta.
You can choose to register for either one or both the workshops. But spots are limited. So please apply in advance.
Session I: Your LLM Stack from Prototype to Production - design considerations and best practices for production-grade LLMOps for RAG applications, fine-tuning models, and more by MLOps Community.
Description: If you've been building prototypes of LLM-powered applications like RAG or fine-tuning your LLMs, you might be wondering what your LLM stack should look like as you transition from basic prototypes to sophisticated, production-grade systems. This workshop is designed to guide you through the critical aspects of developing and deploying robust LLM applications. We'll explore system design trade-offs, discuss best practices in LLMOps, and delve into design patterns for creating robust pipelines for RAG agents, fine-tuning your LLMs, and more.
Instructor: Rahul Parundekar has been building AI-powered systems for 15+ years. He has worked building autonomous agents for Toyota, researched knowledge graphs, and led ML Engineering and Ops teams in building a system that serves over 38 models of different modalities. Most recently, Rahul has architected and built the MLOps platform for Adept.ai, GenAI prediction pipelines for Simplified.com, and PromptOps for aihero.studio.
Format: Code walkthroughs and architecture diagrams (unfortunately, we won't have time for everyone to run code)
Who is it for: This workshop is ideal for AI/ML engineers, product managers, engineering/product leaders, and anyone who has built some prototypes and looking to understand best practices for a production-grade LLMOps stack.
Key Takeaways/What You Will Learn: This session will cover a breadth of information with code walkthroughs on:
Best Practices in LLMOps: Learn the fundamentals and advanced techniques for efficient LLM operations.
System Architecture and Design: Understand how to design scalable and robust system architectures for LLM applications.
Fine-Tuning LLMs: Gain practical knowledge on fine-tuning techniques outside of the Jupyter Notebook.
Production-Grade Model Serving: Discover strategies to transition from prototype to serving LLM models in production.
Prerequisites:
Prior experience with basic fine-tuning or serving LLMs, RAG, or similar technologies is beneficial but not required.
Our code walk-through examples will be in Kubernetes, but we'll keep it abstract enough that the design patterns are transferrable.
Resources: We encourage you to come in having built an RAG prototype or fine-tuned a model in a Jupyter Notebook.
Session II: GenAI Application Prototype to Production: From tinkering with GPT to Shipping GenAI apps with high-quality and low-cost by Verta.
Description: Tinkering around with GPT is easy but building a high-quality app takes weeks: prompting models “just right” is hard, evaluating the results of your model + prompt on real data is challenging, and systematically choosing the best model from the ever-changing menagerie of open-source and proprietary models is even harder.
Join the Verta team for a hands-on workshop to learn about how to avoid spending days in prompt and model hell and set up a systematic experiment-evaluate-refine workflow to take your idea from a prototype to a real product you can share with the world.
Format: Hands-on.
Who it is for: This workshop is for Software Engineers building AI Products (aka AI Engineers), Product Managers, and other GenAI builders who want to build or are already building an LLM application.
What you will learn: This is a hands-on workshop that will take you from just experimenting with GPT to building and deploying a real LLM application of your choice using the best practices in model selection, prompting, and evaluation.
Attendees will learn how to:
Assess and identify the best model for your app, whether open-source or proprietary
If switching from GPT to an open-source model, learn about best practices for picking OSS models
Perform effective prompt engineering and assess the quality of your results
Build and deploy end-user GenAI apps
Prerequisites:
An idea for a GenAI app you want to build/are building.
Some familiarity with prompting and GenAI models.
Excitement for experimentation and app building to take your idea to production!
**PLEASE BRING YOUR OWN LAPTOPS**
Resources:
Access to Verta’s GenAI Workbench (will be provided to all attendees for the course).
Slides and additional materials will be shared at the workshop.
Thanks to the amazing folks at Microsoft Reactor for being a community partner and hosting our upcoming events!!
Note: To comply with the venue, we've added a few questions that have been requested of us. Thank you for understanding.