MLOps Song?

Adam and Demetrios sat down with Mohamed Elgendy the CEO of Kolena to talk about testing-centric development in ML. Mohamed who is ex-amazon, ex-twillio, and the author of "Deep learning vision systems", schooled these two on testing.

The conversation goes so much deeper than just software testing and unit testing. Mohamed explains why it's much more advantageous to look at testing with a wholist mindset.

"The last thing any engineer wants to hear is you need to write more tests, so how can we fix that?" - Mohamed Elgendy

Neoway is a Brazilian company that generates intelligence for decision-making and productivity for the processes of marketing, compliance, fraud prevention, legal analysis, and credit management. Since its foundation, the company has invested in a data platform for collection, ingestion, treatment, and presentation that can be used to create products and services for different use cases.

In the last five years, data science and machine learning have emerged to bring more intelligence in solving customer problems and currently play a key role in the solutions that make up the portfolio. Models, indicators, intelligent journeys are examples of data science-powered applications delivered on our web application or via APIs.

The company’s growth brought new challenges to maintain a good communication structure among teams and fast business value delivery to the customers.

MUST-LISTEN EPISODE. NOT KIDDING.

This week, we had the one and only Shreya Shankar! Many of you have suggested her as a guest to have on, and we finally did it. You’ve definitely seen her writing about how to build machine learning systems on Twitter and many other places.

Her reflections are essential reading for any machine learning professional, especially her latest series on machine learning monitoring. With experienced at Google, Stanford, Viaduct, and Amplify Partners, Shreya is a legit ML superstar who dropped knowledge and hot takes all through this episode. It’s one I’ll personally be re-listening to myself.

Our conversation was split into four parts: Shreya’s lessons on ML systems dev from her past professional experiences, what she thinks about ML monitoring, how she creates such authentically interesting technical content, and a lightning round. Shreya delivered an extremely interesting hot take about how to ACTUALLY do machine learning.

No spoilers here, so you have to listen to find out. I will also say that her advice on what ML engineers should focus on learning in 2022 was super useful for me personally, check this episode for that alone.

- Vishnu

Some cool things have been happening recently. Looks like people are organizing meetings within the different channels with various goals.

Last week we had a round table discussion equipped with breakout rooms for the 20 of us that wanted to discuss different stacks and MLOps design choices. More info about that initiative in the #pancake-stacks channel on slack.

Another group that is going to be meeting this week is the Women of MLOps. It's happening today and you can still join if you are interested! Check out the public calendar for the link to the virtual conference room.

We are doing a new type of meetup this week! Hamza will be leading a workshop on how to build an ML platform with ZenML. It's scheduled to go 90 minutes and it will be very programing heavy. We expect people to follow along with us while building the stack.

Expect to create a pipeline with scheduled orchestration, drift detection, and automatic metadata tracking for reproducibility. Hamza will showcase how ZenML helps you deploy the pipeline using your favorite tools such as MLFlow, Kubeflow, Evidently, and more.

Sub to our public calendar to make sure you never miss an event.