DataTalks.Club

From Scratch to Success: Building an MLOps Team and ML Platform - Simon Stiebellehner

Jun 30, 2023
Ask episode
Chapters
Transcript
Episode notes
1
Introduction
00:00 • 2min
2
MLOps and ML Platforms: A Career Journey
01:37 • 3min
3
The Importance of Machine Learning Operations
04:11 • 2min
4
How to Build a Platform to Serve Hundreds of Models
06:10 • 3min
5
The Importance of Understanding the Data Science Workflow
09:23 • 2min
6
The Importance of Knowledge in a Team
11:09 • 2min
7
How to Build a Good ML Platform Team
13:17 • 2min
8
When to Start Building a ML Platform
15:06 • 3min
9
How to Build a Platform for Data Science Workflow
17:39 • 4min
10
How to Prioritize Your Batch Processing Platform
21:28 • 2min
11
The Advantages and Disadvantages of a Data Processing Platform
23:29 • 2min
12
How to Build an Experiment Track for Your Data Scientist
25:33 • 2min
13
The Importance of Model Registry in Data Science Platforms
27:13 • 3min
14
How to Build an Experiment Tracker From Scratch
30:22 • 3min
15
The Importance of Thin Layers in SageMaker
33:16 • 2min
16
Building a Platform for Sensitive Use Cases
35:20 • 2min
17
How SageMaker Can Help You With Data Governance
37:24 • 2min
18
Data Governance for ML Platforms
38:59 • 1min
19
How to Store Your Data for GDPR Compliance
40:29 • 2min
20
How to Draft an Infrastructure for a Profit-Oriented Organization
42:04 • 2min
21
Building a Platform in Parallel to a Business Case
43:36 • 2min
22
How to Build an Abstrativity Platform for Your MLOps Team
45:33 • 3min
23
API Design for MLOps
48:17 • 2min
24
The Problem With MLOps and Machine Learning Engineering
50:24 • 3min