MLOps Double Shot

Yup, that happened

The latest conference/meetup was a blast! Thanks to everyone who came out and hung out with me. I learned a ton from Karan Goel who gave the first talk. I love his notion of where a system should be touched by human and what parts shouldn't be. Made me think of how many times I hear people say human in the loop, but which part of the loop? And is that the best part of the loop for humans to be? Also made me think of all those times machines have touched me and I never said anything.

You know I couldn't miss an opportunity for me to play the guitar and improvise lyrics. I've taken the liberty of queuing that part up for you in case you want to see it. Shout out to the person who suggested lyrics about sheep and goats like 4 times. I'm thinking about starting a cameo.com page what do you think about that?

Also worth taking a moment right now to thank the good people at Tecton for organizing the event. And for being a sponsor of this glorious community!

More MLOps and Security

You guys have heard us ranting and raving about the different security challenges posed by ML systems. We're going deeper!

What was the session about? Many of you listened when Diego joined us a few weeks ago to talk about some of the more high level challenges of security in MLOps. Sahbi took us a level deeper and introduced to specific security challenges, as well as new tools from companies like Microsoft that are helping solve MLOps security challenges.

Who is this guy? For this coffee session, Demetrios and I spoke with Sahbi Chaebi, a senior data scientist at SAS about security in MLOps. Sahbi is one of the most pleasant people we've ever had on the podcast, and we had a great discussion!

Till next time,
Vishnu

MLOps Standards at Intuit

For this coffee session, Vishnu and I spoke with two very wise men about standardizing the MLOps space. It is a dream many are thinking about these days from my old boss Luke Marsden, with the work he is doing in the #mlops-stacks channel, to the AI infrastructure Alliance led by Dan Jeffries!

What was the session about? Well, with the explosion in tools and opinionated frameworks for machine learning, it's very hard to define standards and best practices for MLOps and ML platforms. Based on their building AWS SageMaker and Intuit's ML Platform respectively, Alex Chung and Srivathsan Canchi spoke with us about their experience navigating "tooling sprawl". They discussed their efforts to solve this problem organizationally with Social Good Technologies and technically with mlctl, the control plane for MLOps.

Who are these guys? Alex is a former Senior Product Manager at AWS Sagemaker and an ML Data Strategy and Ops lead at Facebook. He's passionate about the interoperability of MLOps tooling for enterprises as an avenue to accelerate the industry.

Srivathsan leads the machine learning platform engineering team at Intuit. The ML platform includes real-time distributed featurization, scoring, and feedback loops. He has a breadth of experience building high scale mission-critical platforms. Srivathsan also has extensive experience with K8s at Intuit and previously at eBay, where his team was responsible for building a PaaS on top of K8s and OpenStack.

If you are interested in this stuff there is a channel #sgt in slack to stay up to date!

But I like It

I first heard about Ewan in 2019 cause a colleague of his at the BBC told me he was the one to talk to when it came to ML infra and tooling.

Well, I lost touch with him since 2019 frankly cause he never responded to my cold linkedin sales messages, but he was brought back onto my radar after another colleague of his from when he worked at Skyscanner told me we oughtta have him on.

This time when I reached out he responded! In typical British fashion, Ewan told me I don't know what MLOps is, but I think I'm starting to like it. He now heads up the data science team at a little company called Forcast.

So what to expect from this meetup?
A data/ML system in production is different from both traditional software engineering and traditional data science/analytics workflows.

These differences can be pretty subtle, and trying to use your traditional skill sets to solve these new problems doesn’t work.

Ewan will demonstrate a realistic machine learning system in production and use this demo to show some patterns that he has found useful for living with ML in production, and maybe debunk a couple of myths along the way.

Ewan will also point at some MLOps developments that he really likes, and show some things on his wish list.

See you tomorrow at the usual time 9am PST/5pm BST. Pro-tip, we have a public cal you can subscribe to.