Scribd
Upload
7 views25 pages

MLOps Overview by Folkert Stijnman

MLOps is a set of practices aimed at designing, deploying, and maintaining machine learning models in production efficiently and reliably. It encompasses the entire machine learning lifecycle, including design, development, and deployment phases, while improving collaboration and automating processes. Various roles, both business and technical, are involved in MLOps to ensure successful implementation and performance monitoring.

Uploaded by

asma.belkahla.chaabane
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
7 views25 pages

MLOps Overview by Folkert Stijnman

MLOps is a set of practices aimed at designing, deploying, and maintaining machine learning models in production efficiently and reliably. It encompasses the entire machine learning lifecycle, including design, development, and deployment phases, while improving collaboration and automating processes. Various roles, both business and technical, are involved in MLOps to ensure successful implementation and performance monitoring.

Uploaded by

asma.belkahla.chaabane
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

What is MLOps?

MLOPS CONCEPTS

Folkert Stijnman
ML Engineer
Machine Learning Operations
...is the set of practices to design, deploy and maintain machine learning in production
continuously, reliably, and efficiently.

Focus on machine learning 'in production'

The full machine learning lifecycle

MLOPS CONCEPTS
The origin of MLOps

Practices and tools to deliver software applications

Development and Operations used to be separate

MLOPS CONCEPTS
Why MLOps

1 [Link]
learning

MLOPS CONCEPTS
Why MLOps
Improved collaboration

Automation of deployment

Monitoring of model performance

MLOPS CONCEPTS
Let's practice!
MLOPS CONCEPTS
Different phases in
MLOps
MLOPS CONCEPTS

Folkert Stijnman
ML Engineer
MLOps lifecycle

MLOPS CONCEPTS
Why the machine learning lifecycle?

Structures the ML process

Defines key players at each stage

Toolkit for optimization

MLOPS CONCEPTS
Design phase

Context of the problem

Added value
Business requirements

Key metrics

Data processing

MLOPS CONCEPTS
Development phase

Develop machine learning model

Experiment with data, algorithms, and


hyperparameters

Model ready for deployment

MLOPS CONCEPTS
Deployment phase

Integrate the machine learning model in


business

Deploying the model in production

Monitoring the performance

MLOPS CONCEPTS
MLOps lifecycle

MLOPS CONCEPTS
Let's practice!
MLOPS CONCEPTS
Roles in MLOps
MLOPS CONCEPTS

Folkert Stijnman
ML Engineer
Machine learning lifecycle

Business roles

Technical roles

MLOPS CONCEPTS
Business roles
Business stakeholder
Subject matter expert

MLOPS CONCEPTS
Business roles: business stakeholder

Budget decisions

Vision of company
Involved throughout the lifecycle

MLOPS CONCEPTS
Business roles: subject matter expert

Domain knowledge

Involved throughout the lifecycle


Interpret and validate data

MLOPS CONCEPTS
Technical roles
Data scientist
Data engineer

ML engineer

MLOPS CONCEPTS
Technical roles: data scientist

Data analysis

Model training and evaluation

MLOPS CONCEPTS
Technical roles: data engineer

Collecting, storing, and processing data

Check and maintain data quality

MLOPS CONCEPTS
Technical roles: ML engineer

Versatile role

Specifically designed for complete machine learning lifecycle

MLOPS CONCEPTS
Additional roles involved in ML
Data analyst, developer, software engineer,
backend engineer

Responsbility of roles can vary depending


on application of machine learning

Startup is different from a large enterprise

MLOPS CONCEPTS
Let's practice!
MLOPS CONCEPTS

You might also like