DVC Community September Updates

???? Hi friend! Hope you enjoyed your summer!? Let's jump into the latest!

?The Latest

??? New Blog Posts

We have some exciting new content on the blog:

• (Tutorial) Leveraging LLMs in Chabots: The DVC Approach by Ryan Turner - Looking to design efficient and organized chatbots? This example shows you how to do that using a RAG chatbot.
• (Tutorial)?Automate model deployment to Amazon SageMaker with the DVC Model Registry from Tapa Dipti Sitaula and David de la Iglesia Castro
• (Tutorial)?Fine-tuning Large Language Models with a Production-Grade Pipeline by Alex K.
• (Company News)?Achieving SOC 2 Type 2 Compliance - Our Continued Commitment to Data Security - the team is hitting new daily milestones!

?? New videos

Community Spotlights

• DVC in Machine Learning Engineering and MLOps Practices - Mikhail Rozhkov presents at DSC Adria Conference 2023
• How to Choose a Vector Database in 2023 - Noé Achache .? If you missed our meetup, this is the edited version!
• Empowering Fast and Reproducible Machine Learning - Denis Stalz-John at Dida Conference on his workflow with DVC, Hydra, Kubernetes, and Streamlit

Product Updates

• DVC Extension for VS Code -?Shareable DVC Studio Machine Learning Experiment Links from Matt Seddon
• DVC Extension for VS Code - Zoom and Pan on Focused Plots also from Matt (This one's a game-changer!)

?? What we're looking at

• Comparing and Explaining Diffusion Models in HuggingFace Diffusers - if you are looking for good resources on this topic, Mario Larcher does not disappoint!

You all know that GitOps is integral to our philosophy, so we thought you'd enjoy these:?

• If only all typos were as fun - https://guthib.com/
• Working on your Git skills?? Here's a great resource for learning Git with cats ?from Tomomi Imura !

"Do what you can, with what you have, where you are."? Theodore Roosevelt

??What's Coming!

• MLOps Community Meetup - IN PERSON - Seattle - October 19th. This will be a Happy Hour PLUS great presentations from Diego Kiedanski, PhD on Implementing MLOps in AI-Driven Organizations and Yujian Tang on vector databases for LLMs and another topic from Pendulum on LLMs.
• MLOps World/Generative AI Summit in?Austin, TX October 25-26 in person with a virtual component on the 24th. If you are coming, please sign up for the ???MLOps and Hops Brewery Hop event PRE-CONFERENCE on the 24th from 5-7pm, and then come say hello at Booth 5!

?? Community-Generated Content

OpenGeoHub Summer School with Anita Graser: DVC and Moving Pandas for Geospatial Data - Anita Graser ?? gives a presentation/workshop on moving Pandas and DVC for geospatial data for Open GeoHub.? She uses DVC for a data version control use case to cut down on time to run experiments and the ability to accurately reproduce, data, and experiments.

Introduction to NAVER Place AI Development Team - yunsang ju gives an in-depth view into their CV and NLP workflows for the apps they are building for Naver Place.? They leverage DVC for data and model management in this use case.?

DVC - Great Tool for Data Science - Dmitry Nesmeyanov shares a tutorial in Russian language (or just use Translate this Page on Mac) on DVC and the extension for VS Code.? His team at LOCKO-Bank uses DVC and the extension for data version control, model versioning, and experiment tracking.?

Building Better ML Systems ?-- Chapter 3: Modeling, Let the fun begin - Olga Chernystska mentions DVC for data version control in this great piece in a three-part series on modeling.? The article is full of good advice and other great resources.

Mastering MLOps: A Definitive Guide for AI Product Managers - Nicholas Dass shares this LinkedIn article that advocates for DVC use for data version control and pipelines.? It's a good article with a discussion of ML teams' maturity levels, how to assess that, and great questions to ask yourself about your situation to determine your next steps.

The Evolution of Software Development: MLOps, CI/CD, and Modern Practices - Parvez Sohail proposes a distinction between AIOps and MLOps in this interesting post that also discusses maturity levels and DVC and CML's use in different workflows.?

Trying to Turn Machine Learning into Value - This "oldie" but goodie came across my screen again this week from Ricardo Manh?es Savii .? In it, he created a great diagram to explain the CD4ML process.? We had a discussion at the time about how it can be challenging to share these detailed diagrams because they are difficult to read in a blog post, often even when expanded.? We also discussed the idea of creating a kind of crowd-sourced MLOps Architecture Playground in Miro, where the Community could share their images and get feedback from others.? this week he shared with me his updated version in Miro and it's beautiful! ?? See Ricardo's detailed diagram here! And thank him in Discord here !

?Community Question: Would you be interested in participating in a crowd-sourced MLOPs Architecture playground that provides you the tools to create MLOps architecture diagrams to share with others and be able to view and discuss those of others?

? Yes, absolutely I would love to have a place to work on and share my MLOps Architecture diagrams.

? Yes, I would love to be able to view others MLOps Architecture diagrams in more detail and learn from others.

?? Even if I want to, I wouldn't have permission to share my company's diagrams.

?? It's a cool idea, but I'm not sure I would have time.

I loved hearing recently from some Community members who are speaking about DVC and their use cases with it around the world!? Please reach out to me here in LinkedIn or @jendifig in Discord and let me know about these.? If there are recordings I'd love to share them on our YouTube channel and showcase your great work!

???? See you in October!

To your continued success,

Jeny

Community Manager?

Subscribe to this monthly newsletter here.