Sages and Saints

Sagify

The Story: In late 2017 my greek brother Pavlos Mitsoulis was a data scientist suffering through the painful experience of configuring his own EC2 instances in order to train and deploy ML models. Fraught with distress he tilted his head upwards, threw his hands in the sky, and clamored "there must be a better way!" Sadly the state of MLOps was nowhere near as booming as it is today, so with the fire of determination burning inside of him he set out to make things right for all other data scientists who found themselves in the same predicament.

The Solution: The idea was simple and some even laughed at his naivete. Train and deploy models by implementing 2 functions on AWS? "This is blasphemous!" They responded and even created custom slack emojis to call him out. None the less Pavlos persisted. Train() and predict(), train() and predict(), train() and predict(). Two functions was all he needed.

Enter Sagemaker: As Pavlos became more and more enamored with the idea of making life for himself and fellow data scientists easier, AWS did something unexpected; they released Sagemaker. "Oh no, they got there first" he thought to himself. Then it dawned on him, Sagemaker is an ML engine, not an ML platform. Realizing he could take the reins and stand on the shoulders of the beast this would help him get to his solution sooner! Train() and predict(), train() and predict() like an incantation fueling his late night coding sessions.

Sagify Is Born: After numerous sleepless nights and nearing defeat countless times Pavlos rose like the phoenix from the ashes with an easy to use CLI MLOps tool in hand. Sagify is for all those data scientists who feel their voice falls upon deaf ears. For those data scientists who only want to focus 100% on ML; just training, tuning, and deploying models. For those that are already on AWS, a star is born to leverage Sagemaker as a backend ML engine, so they can work smarter, not harder.

*This is based on a true story, some creative liberties have been taken for the sake of entertainment. :)

How I Learned To Stop Worrying And Love The Bomb

Igor Lushchyk, a data engineer at Adyen, rolled through our meetup last week and broke down some of the most important nuggets of wisdom he has picked up over the last 5 years having his head down in the MLOps world.

Who said anything about suffering? One surefire way to suffer according to Igor is by trying to implement heavy systems without knowing how they work. Better instead is to start with something lightweight and learn the inner working of what's going on behind the scenes that way if anything breaks you will have a much better time debugging.

+Props to another solid meetup. Check out the video here and the podcast here. Also we have a short clip from the meetup here

Simple Rules for Domain-Specific Model Monitoring

Deploying models as services is all the rage nowadays. Monitoring is a core part of service deployment, but how does it translate to machine learning models? Also, given how specific machine learning models can be, how do you monitor with the right context?

Enter Domain-Specific Model Monitoring! Community member Lina Weichbrodt, a lead machine learning engineer at DKB Bank and former tech lead at Zalando, wrote a great blog post about how to solve these problems. I don't want to give away her secrets, but at a high level, it's all about understanding user behavior and designing heuristics based on their expectations and needs. As ML professionals, it's very easy to get caught up in performance metrics, etc. Lina does a great job of showing us how to move away from that mindset with specific tactical tips.

Operationalizing Machine Learning at a Large Financial Institution

Daniel, our guest this week told me he wasn't sure people would want to hear about his use case because of the high percentage of legacy systems that they still use at his company. That is precisely why we need to have him on! Not everyone has the luxury of being able to use k8s or the next hottest toy at work due to all kinds of reasons. This one is for you.

Meetup: The Data Science practice has evolved significantly at Regions, with a corresponding need to scale and operationalize machine learning models. Additionally, highly regulated industries such as finance require heightened focus on reproducibility, documentation, and model controls.

We plan on talking with Daniel about how the Regions team designed and scaled their data science platform using devops and mlops practices. This has allowed Regions to meet the increased demand for machine learning while embedding controls throughout the model lifecycle. In the 2 years since the data science platform has been onboarded, 100% of data products have been successfully operationalized. (Take that Gartner and your damn 80% of models never get deployed statistic!)

Bio: Daniel Stahl leads the ML platform team at Regions Bank, and is responsible for tooling, data engineering, and process development to make operationalizing models easy, safe, and compliant for Data Scientists.

Daniel has spent his career in financial services and has developed novel methods for computing tail risk in both credit risk and operational risk, resulting in peer-reviewed publications in the Journal of Credit Risk and the Journal of Operational Risk. Daniel has a Masters in Mathematical Finance from the University of North Carolina Charlotte.

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