CI/CD/CT/MLOps

Back to Basics

Kseniia gave us a back to basics presentation about covering what she called the foundations of the ML lifecycle.

These days many of the common mistakes she has seen can easily be remedied by different tools mentioned. She did agree though, there has yet to be a tool that can substitute the P0 need for clear communication.

Special shoutout to the diagram she showed towards the end which breaks down the different stakeholders and their needs. Oh yeah, and I did a guitar intro which I now think I am going to start doing more often. Let me know if you think that is a good idea or a great idea!

Full convo is out on youtube and in podcast land.

Mon Ray The GitLab Way

It's an eternal set of questions in our community: DevOps or MLOps? Which makes more sense to organize things around? How does automation work in the context of ML? We've covered a topics a fair amount, and we went even deeper this time with the pleasure of Mon Ray's participation. Mon is a solutions architect at Gitlab, one of the world's foremost DevOps companies She shared her lessons from experience working on MLOps with her clients, as well as her crossfunctional background as a data scientist and consultant.

Mon brought a great sense of empathy to the table, which is a crucial part of figuring technical solutions that actually stick. After all, people need to be bought in on a personal level to the solutions they are a part of. One of my favorite points that Mon made is around YAML literacy and how foreign that can feel for many data and machine learning professionals, despite its ubiquity in software engineering and DevOps contexts. A simple obstacle like YAML literacy can pose a large mental challenge to embedding DevOps priniciples like CI/CD into the machine learning workflow. As someone that has recently started to get more comfortable with YAML and is teaching others, I totally agreed with Mon about this "small" tactical challenge and the big problems it poses.

Check out the video and podcast to get more great lessons on practical ways to integrate MLOps principles into your workflow from Mon!

Till next time,
Vishnu

Part 2

So far we have spoken with half the authors to the new book Practical MLOps. That is all about to change tomorrow. Alfredo will be joining us to talk about the importance of bringing DevOps principles and discipline into Machine Learning.

Alfredo and I will make sure to highlight the process of creating the MLOps role and what is included. rest assured we will touch on automation, constant feedback loops, and the number one objective - to ship models into production.

Additionally, we'll cover some aspects of getting started with Machine Learning that are critical, in particular how democratization ML knowledge is critical to a better environment, from libraries to courses, to production results. Spreading the knowledge is key!

Who knows, we might even give away some books too!

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.