🐤 Feature stores and CI/CD for machine learning with Qwak.ai VP Engineering, Ran Romano
Aug 11, 2021
auto_awesome
Ran Romano, VP Engineering at Qwak.ai, discusses the evolution of job titles in machine learning, challenges of using Jupyter notebooks in production, and the importance of adopting a CI/CD approach. They also talk about the challenges in scaling ML models to production, ensuring data reproducibility, and using open source solutions in their ML platform.
Effective CI/CD processes are crucial for deploying machine learning models into production and ensuring replicability and versioning of models and data pipelines.
A feature store plays a vital role in managing reproducibility of training data sets, allowing for easy sharing and reuse of features across different machine learning models.
Deep dives
Building an MLOps Platform
Ron discusses the motivation behind starting the podcast, which is to bridge the gap in knowledge and best practices between different machine learning teams. He believes that the information about bringing machine learning projects into production is not widespread enough, and many people are not familiar with the best practices and approaches. This podcast aims to interview professionals working in various machine learning teams to understand how they are successfully deploying their projects.
Defining Machine Learning Engineering
Ron explains the ambiguity surrounding the roles in the field of machine learning, specifically the distinction between data scientists, data engineers, and machine learning engineers. He shares his experience of building a machine learning platform at Wix, where the team referred to themselves as 'machine learning infrastructure engineers'. Ron believes that the trend is moving towards separate roles for data scientists and machine learning engineers, but acknowledges that some organizations may still prefer a more integrated approach.
The Importance of CI/CD in Machine Learning
Ron highlights the significance of effective CI/CD (Continuous Integration/Continuous Deployment) processes in the field of machine learning. He discusses the challenges his team faced in automating the deployment of models into production at Wix. Ron emphasizes the need for replicability and versioning of models and data pipelines and shares how they built a CI system that allowed data scientists to independently deploy their models.
Building a Feature Store
Ron explains the need for a feature store in machine learning platforms and its role in managing the reproducibility of training data sets. He discusses the challenges of feature extraction and integration with data sources. Ron describes how they built a feature store at Wix, emphasizing the importance of having a single, curated, and discoverable source of truth for features. He explains how the feature store facilitated sharing and reuse of features across different machine learning models.
In this episode, I'm speaking with Ran Romano from Qwak.ai. Ran built the ML platform at Wix, and we discuss the various data roles, when organizations should focus on ML infrastructure, solving the hard problems of features stores, and one approach to building an end-to-end ML platform.
---
Timestamps:
00:00 Podcast intro
01:00 Guest intro
01:30 Getting into the world of ML and ML Engineering
02:25 The line between Data Engineer, ML Engineer, and Data Scientist
03:50 The future of data roles – what are the trends?
07:21 The most exciting part about taking ML models into production
09:45 Jupyter notebooks in production (again??)
10:41 Signs that notebook productionization might not work
11:42 Building ML-focused CI/CD systems
15:32 Early days of building out the Wix ML platform
16:22 Signs that you might need to focus on ML infrastructure in your organization, and how to convince other stakeholders.
19:21 What part of the platform that you built are you most proud of?
23:51 Defining a feature store and the training/serving skew
27:24 Onboarding data scientists to using a feature store
33:49 When is it too early to build an ML platform?
35:33 Open source components – What parts of your platform did you choose not to build yourself?
40:16 Qwak.ai – What are you working on currently?
41:07 How do you define an "end-to-end" platform in the case of Qwak
44:25 End-to-end vs. Integrated – Advantages and disadvantages
---
Relevant Links:
- Qwak.ai: https://www.qwak.ai
- Wix ML Platform presentation by Ran: https://www.youtube.com/watch?v=E8839ENL-WY