DVC Community January Updates!

???? Hi friend! Happy New Year!? 2023 was a whirlwind of change.? Let's see what 2024 has in store, and buckle up because I've got a long list of awesome Community Content at the end!

?The Latest

?? New Videos

? ? ? Tutorials/Conference Talks

• GitOps Best Practices: DVC Studio Model Registry for ML Development - Jelle Bouwman gives us a thorough tutorial on how to get started,?use, and automate stages in?the DVC Studio Model Registry for production.
• How to NOT get ML Models into Production - Ryan Turner's lightning talk a MLOps World and Generative AI Summit.? This is a satirical look at the subject with some helpful tips!

? ? ? ?Product Updates

• DVC + Git LFS for Seamless Hugging Face Model and Dataset Versioning - Peter Rowlands shows of this MUCH anticipated feature.? The integration not only works for Huggung Face models but with anything that is using Git LFS for Storage. (see corresponding Blog post below.)
• New Way to Access ML Model Files in DVC Studio - Jelle Bouwman shows the new way to access your mode files whether they live in S3, Azure, or GCS.
• DVC Studio: Shift Select Feature added for bulk actions - Jelle Bouwman shows how easy it is to show, hide or delete experiments in your experiment table?
• How to Connect to DVC Studio from the DVC Extension for VS Code - Julie Galvan shows how easy it?is to collect your access token and connect the work from your local VS Code environment to your team in DVC Studio.
• How to Debug Plot Errors in the DVC Extension for VS Code - Julie Galvan 's back with this great feature that makes it much easier for you to identify faulty plot code!

??? New Blog posts

• Integrating DVC and Git LFS via libgit2 filters - Peter Rowlands explains in this post how he and the team accomplished the recent integration between DVC and Git LFS.
• Tutorial: Automate Data Validation and Model Monitoring Pipelines with DVC and Evidently - Mikhail Rozhkov wrote an in-depth tutorial on using DVC and Evidently together to ensure your models remain precise and verify that your data consistently reflects the real-world scenario.

?? What we're looking at - Hint: DATA QUALITY

• 4 Reasons AI in 2024 is On an Exponential: Data, Mamba, and More - I enjoyed this video from AI Explained which was packed with information and links to resources, but especially the quote from Tri Dao, one of the architects of Mamba, who said that "all the architecture stuff is fun...but it seems like the only thing that changes the slope [of the scaling law curve] is the data quality."

[Join the conversations in the following social media posts by clicking on the image.]

??What's Coming!

If you are a Data Scientist, Data Engineer, or ML Engineer, here are some upcoming PyCon conferences that you may want to attend or present at:

• PyCon Namibia - March 4-7
• PyCon Lithuania - April 2-6
• PyCon US - May 15 - 19
• PyCon Italy - May 22-25
• PyCon DE & PyData Berlin - April 22 - 24
• PyData London - June 14-16, Calls for Proposal Deadline: March 4th
• PyCon Estonia - September 5-6, Call for Papers Deadline: April 15th

Next Meetups

• February 14th we will host Charles Vardeman , Computational Scientist and Assistant Professor at the Center for Research and Computing at Notre Dame will share his his work on developing safe, secure, and Trustworthy AI Frameworks highlighted in this post in KD Nuggets.?
• February 21st - Ryan Turner will present on DVCx applications for LLMs.?

Keep your eyes on LinkedIn and Twitter for upcoming info on these events!

?? Community-Generated Content

MLOps Simplified: Mastering CIFAR-10 on Your MacBook - Abhijoy Sarkar shares this article on an End-to-end? MLOps pipeline for machine learning using the CIFAR-10 dataset, DVC is used for data and model versioning in Google Cloud Storage, CI/CD with GH actions, packaging with Docker, inference with FastAPI, and efficient runtime with ONNX.

Advent of Code for Open Source - Yujian Tang and the team at Zilliz put together a wonderful plan for building your skills in a variety of open-source tools over 24 days in December.? You can find all the info in the link. DVCpaticipated on Day 10 and you can find our recommended projects with DVC Live here.?

Optimizing Machine Learning? Workflows: Containerization, Versioning with CI/CD automation - This article by Abjulsamod Azeez discusses optimizing machine learning workflows by incorporating containerization, versioning with DVC, and CI/CD automation. DVC is used for models and data in this approach making these things shareable, experiments reproducible, and tracking versions effective.

Exploring Open-Source Alternatives to Landing AI for Robust MLOps - This article by Anthony C. in Dev.to has a bunch of good tips and tools mentioned in it including a link to awesome-production-machine learning and a Map of GitHub! ??

Building a Generative AI Platform - from NexaStack , this article?discusses the practical use of Kubernetes in addressing challenges in adapting Artificial Intelligence, focusing on machine learning and deep learning techniques. It emphasizes the importance of Kubernetes in managing containerized architectures for scalable AI models in production. DVC is used for versioning data models, directories, and intermediate results.

A Look Inside the Cloudflare MLOps Platform - Keith Adler and Rio Harapan P. write this article that provides insight into Cloudflare's MLOps platform which includes DVC, emphasizing its use of ML and AI in core services such as Web Application Firewall.

Future of Data Science Project Development with OpenAI and Hugging Face API's - Krish Naik gives an overview of what will be important to project development and getting a job with all the changes that have happened in 2023. He is still recommending DVC as basic knowledge you need to have to be able to function successfully.

Open Source Contribution End-to-End Machine Learning Projects in 2024 - Krish Naik also gives this video overview on contributing to Open Source projects.

Traditional Software vs Machine Learning Software - Filipa Castro presents how her team uses DVC as well as other tools in her data science work on smart products at Continental, best known for their?tires.

Top 6 Dataset Version Control tools for your machine Learning Workflow - Raksha leads with DVC as the first in this list of tools and after going over each provides a short tutorial on how to get started with DVC.?

Two Harvard AC215 Final Projects Using DVC

• Using AI to Discover Fixer-Upper Deals - Tony Hua and team write about his experience building their final project to identify fixer-upper homes using DVC to manage the versioning of their data
• SnapNutrition: Bridging the Gap Between Technology and Dietary Well-Being - In this project Christina Wang and team use DVC and other tools in their work to identify nutrition information from images.? That would be so much easier than entering that info item by item! That gets a ?? from DeeVee!

Non-English content?on DVC

• Russian: The end of Childhood - how to stop playing with notebooks and create an ML pipeline.?- In this second talk of three, Alexander Khinevich, Senior Data Analyst at the Center for Financial Technologies shows how to they came to reproducible experiments with DVC.
• Armenian:?Data Streams through DVC - Looking for a tutorial in Armenian??Infraheads provides this one entitled Data Streams with DVC that takes you through all the info on getting Data pipelines set up
• French: Une architecture GitOps pour vos projects ML avec DVC - Bertrand Nau discusses the GitOps methodology, highlighting the advantages of collaboration, reproducibility, and history tracking and how DVC provides these advantages to the machine learning space.

Suffice it to say, 2024 is bound to be packed with more learning and growing.? Let's get at it! ?? ?

To your continued success,

Jeny De Figueiredo

Community Manager, DVC.ai