Current Mood: MLOps Workshop

Author, Speaker, Olympian

Little plot twist in our last meetup. Alfredo Deza the co-author of practical MLOps started out his career as an Olympian from Peru. Things have changed a bit since those days. He has now penned some incredible books including 'Python for DevOps' and his latest 'Practical MLOps'.

We spent the session diving deeper into concepts Alfredo and his co-author Noah present in the newest book. Alfredo also gave us a few war stories and common-sense lessons that we can keep in our back pocket. You know, for a rainy day.

You can never have enough tests - When it comes to Shipping something to production, catch things as early as possible. The closer you catch something to production the more expensive it will be.

There is always something to automate - Even if you think it's all automated already, check again. I asked about the pushback that some like myself give when it comes to automating everything in ML. Alfredo shared a nice war story to iterate his point and put mine to shame!

Full convo is out on youtube and in podcast land.

More Than Words Monitoring

Anyone on the community Slack has surely seen Fiddler.AI's two co-founders, Krishna and Amit, pop in and share delightful lessons about AI regulation, MLOps, and model monitoring. Demetrios and I were lucky enough to spend time chatting with half of the terrific duo--Amit Paka, the Chief Product Officer at Fiddler.

There's a lot of ground our discussion covered, from the operational importance of explainable AI to which industries are surprisingly moving fast to implement MLOps best practices. My favorite part of the discussion was Amit's explanation of model performance management, a framework that Fiddler has developed for understanding how to keep models operational in production. I won't dive too deep here (listen to the coffee session), but let's just say that MPM is a natural way of thinking about how to keep all aspects of machine learning working together in an organized, clear feedback loop. Amit explained these ideas in an engaging way with industry examples.

Check out the video and podcast to join us on this awesome deep dive!

Till next time,
Vishnu

Workshop

We have a special meetup this week Building an ML Platform from Scratch. This will be a Live Coding Session.

What to expect?
This workshop will teach you how to set up an ML platform based on open-source tools like Cookiecutter, DVC, MLFlow, FastAPI, GitHub Actions and more.

We'll explain each tool in an intuitive way and the problem it solves, and then build a useful platform that combines all of them. All code will be available on GitHub after the workshop, so you'll be able to integrate it into your existing work easily.

Bio: Alon Gubkin is the VP R&D at Aporia. Writing code from the age of 7, Alon has an unhealthy obsession with programming and a huge love for mixing engineering with data science. When he isn't coding, Alon enjoys playing video games and musing about biology.

Side Note: I have started doing musical intros for our guests at the meetups. True Story: if you don't like it you can let me know by throwing a tomato at your computer screen.

See you on Wednesday aka tomorrow at 5pm UK / 9am California. Click the button below to jump into the event, or subscribe to our public google calendar.

ML Test Score Rubric: Everyone loves the highly practical ML Test Score from Google; now community member Jishnu Mohan and his team made it into a convenient online quiz!
Typical DS Workflow?: This thread might set the record for most lively thread in our community's history. Read for a great discussion on how the transition between DS and engineering works at a number of companies. Thanks for kicking it off Dimi!
Review of Google Vertex: Community member Matt Squire took the time to go through the MLOps lifecycle of products on Vertex, which is Google's hot new ML platform. It's a great summary of where the product currently is.
Paradox of Choice: Interesting thread started by community member Andre that discusses DVC, DeepLearning.ai courses, and production data workflows.