New Tool Tuesday

In A League Of Their Own

Something magical happened last Wednesday in our first official meetup panel; I got to sit back and enjoy the show like the rest of the crowd. Well for the most part at least.

Mike, Willem, and David were awesome to chat with because the chemistry was there before we even started! They all know each other so well, they could dig in and ask questions to drive home a point. The three of them share past experiences from the rich history and learnings they all have and I was content to just be a fly on the wall.

+ The combined MLOps clout on this one zoom call was pretty spectacular between Willem who created Feast, Mike who was part of the Michaelangelo/now runs Tecton, and David who you know... co-created that whole Kubeflow thing. Wow. You can see why I just let them converse with each other and soaked up the great insights.

+ Now, I wouldn't say the conversation stayed completely on track around the idea of 2 tools to get you 90% ML operational, though what I will say is the direction we did end up going in was top quality. One of the key takeaways for me was the idea around standardization within the tooling world and moving towards publicly defined contracts that everyone can agree on.

+ David made it very clear that when we say standards or contracts we want to use a small c not creating huge contracts that will be super opinionated and shape much of the landscape. We are talking about small steps that the ecosystem as a whole can say yes to and then bring a more harmonious vibe to the party! What would those small contracts be? David had a few opinions but i'd love to hear yours on the subject.

+ The best quote of the meetup also came from David and this one might have to go on to a shirt when we actually make some swag "MLOps isn't just about tooling, it's a lifestyle" Bam! Put that in yo pipe and smoke it!

Check out the video here and the podcast here

Cookie Cutter On Steriods

I can not take credit for the description of this one, it was all Alexey this time. However, I want to chat for a minute about a new tool that just came through the community from Neal Lathia head of ML at Monzo. The name Neal Lathia might sound familiar if you saw the awesome coffee session we had with him a few months back.

I wanted to get Neal's take on what it is and why he created this tool called Operator (not to be confused with k8s operators). So Mr. Monzo ML told me....

"Working with ML systems today is a bit like living in two worlds. In one world, I'm thinking about the ML model I've trained and how excited I am to ship it. In the other world, I'm reading the AWS docs (and tutorials, blog posts, and stackoverflow questions) about IAM roles and API gateways and how to tie them all together. I'd like to spend less time in that second world, and more time in the first!"

Neal is a proactive dude and his goal was clear. Inspired by a few internal tools he has at Monzo he decided he'd hack together a CLI tool that has two functions:

Create: to get the boilerplate that's needed to get going.
Deploy: to ship that code to a serverless function in AWS or GCP

Although he is still in the lab creating a model store that hopefully will be released to the world soon, he wanted to get this out so the world can have a play and hopefully spend more time hanging out in that first world he spoke about earlier. (Although multi planetary species are ok in case something happens you know like global warming)

Anyway, Lathia finished off our convo by telling me "I've open-sourced operator so that anyone else can give it a spin and see the ✅ emojis pop up while they deploy their serverless in two commands. If you have any feedback, you can find me on the ML Ops community slack!"

Personally I know a few people that have started using it are loving it, (especially the emoji part), but I'd also be really interested to hear what your thoughts are! You can see a quick video of it in action here, otherwise, click the button to go straight to the source.

Straight from the OG Jet Basrawi

This week, Demetrios and I chatted with Jet Basrawi, MLOps Engineer and Technical Strategist at Satalia, an enterprise AI company. Jet is one of the OG MLOps community members. To give you an idea, he attended the first 5 meetups. I don't know how he heard of them back then, or how he has had the patience to stick around. One thing's for sure, it was great to exchange stories, especially since he's listened to so many coffee sessions and called out some episodes better than us!

Management is NOT leadership
I think we all need to listen to this quote more. I made the mistake of calling Jet a manager, and he quickly pointed out the crucial difference between leadership and management. Managers focus on resource husbandry; leaders focus on influence and inspiration. To be clear, this isn't to rip on management at all! I appreciated Jet pointing this out specifically because for a lot of MLOps professionals, leadership is a crucial part of being effective. You need to influence and inspire your team, organization, and company to really make the most out of ML. That doesn't require you to be a formal manager, but it does require you to apply the principles of leadership. Thanks for this nugget Jet!

"The Kung Fu is in the culture of ML"
Jet made a strong case for being the most quotable coffee session guest in a while. He mentioned in the session how the hard part of MLOps isn't the models or the technology, it's really the culture. It's about emphasizing delivery as a value, learning as a point of pride, and experimentation (and its concurrent failures) as an expectation. This is harder to do in companies that aren't software-native (Jet had some great insight on this), but it's possible, and more importantly, necessary to really succeed at ML. The culture part is the real Kung Fu, and the mysterious, unquantifiable asset that differentiates successful MLOps organizations.

Thanks so much to Jet for joining us! Link to Podcast and Video.

- Vishnu

Short Term Value Vs Long Term Ethical Outcomes

The Theme: One of the challenges to the widespread adoption of AI Ethics is not only its integration with MLOps, but the added processes to embed ethical principles will slow and impede Innovation. In this meetup I'll be speaking with Pamela Jasper about ways in which DS and ML teams can adopt Agile practices for Responsible AI.

Bio: Pamela Jasper is a global financial services technology leader with over 30 years of experience developing front office capital markets trading and quantitative risk management systems for investment banks and exchanges in NY, Tokyo, London, and Frankfurt. Pamela developed a proprietary Credit Derivative trading system for Deutsche Bank and a quantitative market risk VaR system for Nomura. Pamela is the CEO of Jasper Consulting Inc, a consulting firm through which she provides advisory and audit services for AI Ethics governance. Based on her experience as a software developer, auditor and model risk program manager, Pamela created an AI Ethics governance framework called FAIR – Framework for AI Risk which was presented at the NeurIPS 2020 AI conference.

+ Along these lines, the Agile for ML theme has been coming up quite often these days which makes me wonder if we are hitting a wall around the use of agile in ML? Is this a practice we are trying to shoehorn into the ML development lifecycle because it's been around long enough that we can all speak its language? Or is it really the most adequate approach to going about tackling problems in this field?

+ This seed was planted in my head by Charlie You when he was on the coffee session 2 weeks ago and mentioned that sprint planning might not be the best thing for ML engineers because of the nature of the beast. And then last week talking with the agile kung fu master Jet my suspicions were further confirmed when he grieved over the dysfunctional beast that agile had spiraled into at most companies.

+ See you at 5pm GMT / 9am PST tomorrow, Wednesday by clicking the link below.