I am on an island in Greece for the next two weeks. Bear with me on while I'm on my beach vacation! As was my mantra in college - C's get degrees.
Past Meetup
ML Does Security
If you are into hearing about how Machine Learning is being used to keep us safe on the internet, give this one a watch. Monika was truly awesome creating a talk that taught me a ton about the current climate around the deployment of ML in cyber security. Monika has a wonderfully interdisciplinary background that makes her insight worth listening to.
This is just part one of our cyber security extravaganza this week. Scroll down for more!
I synthesised the meetup from a few weeks back with Ouz about how you can get unstuck from continual PoC's. Check it out below!
Also did you see the new website?
Just this past week, we saw a major $70M ransomware attack on Kaseya, which crippled thousands of companies. It's almost as if a day doesn't go by without hearing about a cyberattack on a leading MNC. With all these things
happening, doesn't it make sense that we have such a focus on cybersecurity in MLOps?
Diego Oppenheimer, the CEO of Algorithmia, joined us to talk more about hsi phenomenon. Diego was incredibly insightful as he described the basic forms of security threats and risks that can occur in the context of MLOps. We discussed what ML Engineers should know as they work on security related workflows and topics, and where they should reach out to DevSecOps professionals for assistance.
My favorite part of the conversation revolved around how Diego described the key insight of security: it's all about awareness of risk. You don't need to adopt the most stringent protocol for every single product or platform to be secure! You simply need to critically analyze what kind of risks you're taking on when you opt for organizational flexibility or more strict security standards. After all, IT can't be a barrier to innovation. Check out Diego's coffee session to get
schooled on security (and read this blog post too)!
NO MEET UP THIS WEEK Nothing planned for this week as I will hopefully be on the beach drinking frappes. we are back next week though so I'll see you all then!
Regression Testing for ML: Great write up from the folks at Lakera that describes how to apply the software eng testing practice to machine
learning models. Looking forward to their series on testing!
Uber does it again: As always, the Uber engineering team has incredible work to share. This time, they cover their approach to the challenge of CI/CD for machine learning. Well worth a read!
Everything Store?: Thanks to community member Skylar Payne, we had a great discussion on Slack about why so many tools call themselves a "store" for various data elements.
Sagemaker + Kubeflow: This is why the community is amazing! Community members Mamoon, Ashish, and Richard, all in different places, had a great discussion about the pros and cons of Sagemaker and Airflow. Awesome lessons!