Application in Data Science

In the last reading group session we talked with Ben Wilson about the Machine Learning Engineering book.

Here are some important takeaways:

Chapter 2 presents an adoption of the Agile Manifesto principles into a new list for ML project work. Two very impactful ones are:

Get feedback early and often by scheduling demonstrations throughout development
Frequent collaboration with SMEs

A great tip from chapter 9 regarding decisions on where to split functionality and achieve better modularity of your code is to ask yourself, 'How can I test this block of code?' rather than 'What looks nice?'
It's important to have a middle ground created through a team consensus and lead architect that also understands the business when implementing feature engineering code for a feature store - especially in regards to creating good abstractions of possible operations for given features vs letting differenet data scientists upload their feature engineering code without paying close attention to abstract behavior across the feature code.
Use strong collaboration and team work to have a healthy ecosystem of different teams contributing to the feature store.

For this week's Meetup we have Connie Yang, Lead Data Scientist at Pallet.
She has been leading various data science teams working on integrating machine learning, intelligence, and automation into the entire Pallet product and will talk about her experience in how this can be done. She will draw from both experiences in Big Tech and leading the data efforts at an early-stage tech startup.

We will talk about:
- The different types of data science roles and how those roles can be impactful to the business / product
- The skills necessary for each of the various data science roles and how that could influence someone's choice in the type of data scientist they develop themselves into
- Give examples from real-life applications of data science (both successful & unsuccessful)
- How to break into the field of data science

April is Model Serving Month at the AIIA!

Learn how to serve models in simple and complex environments, see the difference between batch and online, sunset old models, use shadow deployments, monitor the models for speed and latency and much, much more. Check out everything you need to know right here!

Sign up for the event and/or vote for which talk you want to see as the one hour demo or lab. And in case you missed the last event, you can also watch a replay of March’s experiment tracking talks right here.

See you there!