AIM301-300 LEVEL

Implement MLOps practices on

AWS

Dr Sokratis Kartakis (he/him) Ajish Palakadan (he/him)

Snr MLOps Specialist Solutions Architect, EMEA Practice Lead, Modern Data Platform
Amazon Web Services Reply

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda

What is MLOps
People, Processes, Technology

Why MLOps
Business Benefits and KPIs

MLOps People & Processes

Team Structure and Personas

MLOps Foundation Roadmap - Technology

MLOps Maturity: Initial, Repeatable, Reliable, Scalable

MLOps Journey with Data Reply

Partnership, MLOps Accelerator and Customer Success Stories

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Customer Challenges, ML Lifecycle and Personas
CONSIDERATIONS & CHALLENGES LEAD TO ML AND OPERATIONS (MLOPS)

Processes

• Culture
• Lack of cross-functional teams
• Priorities & needs (personas)
• Organizational structure
• Skillsets
• Unique aspects of ML lifecycle

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
What is MLOps?
MLOps Definition
MLOps
Machine Learning
& Operations People

The combination of people,

processes, and technology to
productionize ML solutions Technology Processes

efficiently.

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Why MLOps?
Expected Business Value

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation Expected Outcomes
STANDARDIZE OPERATIONS AND INFRASTRUCTURE FOR YOUR DATA SCIENCE

MLOps Expected
Business Goal Technical Metric Before MLOps Business Value
Outcomes
Time to value up to Improve Speed-to-Value
1 Be more efficient in delivery < 3 months
(from idea to production) 12 months by 4x

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation Expected Outcomes
STANDARDIZE OPERATIONS AND INFRASTRUCTURE FOR YOUR DATA SCIENCE

MLOps Expected
Business Goal Technical Metric Before MLOps Business Value
Outcomes
Time to value up to Improve Speed-to-Value
1 Be more efficient in delivery < 3 months
(from idea to production) 12 months by 4x
Time to productionize existing ML use Reduce FTE overhead
2 Simplify route-to-live 3-6 months < 2 weeks
cases in average 8x

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation Expected Outcomes
STANDARDIZE OPERATIONS AND INFRASTRUCTURE FOR YOUR DATA SCIENCE

MLOps Expected
Business Goal Technical Metric Before MLOps Business Value
Outcomes
Time to value up to Improve Speed-to-Value
1 Be more efficient in delivery < 3 months
(from idea to production) 12 months by 4x
Time to productionize existing ML use Reduce FTE overhead
2 Simplify route-to-live 3-6 months < 2 weeks
cases in average 8x
Focus on innovation
Standardize infrastructure, data,
3 % Template driven development n/a > 85% increasing re-usability
& code
by 85%

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation Expected Outcomes
STANDARDIZE OPERATIONS AND INFRASTRUCTURE FOR YOUR DATA SCIENCE

MLOps Expected
Business Goal Technical Metric Before MLOps Business Value
Outcomes
Time to value up to Improve Speed-to-Value
1 Be more efficient in delivery < 3 months
(from idea to production) 12 months by 4x
Time to productionize existing ML use Reduce FTE overhead
2 Simplify route-to-live 3-6 months < 2 weeks
cases in average 8x
Focus on innovation
Standardize infrastructure, data,
3 % Template driven development n/a > 85% increasing re-usability
& code
by 85%
Standardize onboarding of new Time to instantiate a new MLOps Accelerate ML adoption
4 40 days < 1 hours
teams and ML use cases infrastructure & ML projects across all business areas

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation Expected Outcomes
STANDARDIZE OPERATIONS AND INFRASTRUCTURE FOR YOUR DATA SCIENCE

MLOps Expected
Business Goal Technical Metric Before MLOps Business Value
Outcomes
Time to value up to Improve Speed-to-Value
1 Be more efficient in delivery < 3 months
(from idea to production) 12 months by 4x
Time to productionize existing ML use Reduce FTE overhead
2 Simplify route-to-live 3-6 months < 2 weeks
cases in average 8x
Focus on innovation
Standardize infrastructure, data,
3 % Template driven development n/a > 85% increasing re-usability
& code
by 85%
Standardize onboarding of new Time to instantiate a new MLOps Accelerate ML adoption
4 40 days < 1 hours
teams and ML use cases infrastructure & ML projects across all business areas
Execute the ML solutions without Your data is safe in your
5 Ensure high security standards n/a No internet
internet access in a private cloud private cloud

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation Expected Outcomes
STANDARDIZE OPERATIONS AND INFRASTRUCTURE FOR YOUR DATA SCIENCE

MLOps Expected
Business Goal Technical Metric Before MLOps Business Value
Outcomes
Time to value up to Improve Speed-to-Value
1 Be more efficient in delivery < 3 months
(from idea to production) 12 months by 4x
Time to productionize existing ML use Reduce FTE overhead
2 Simplify route-to-live 3-6 months < 2 weeks
cases in average 8x
Focus on innovation
Standardize infrastructure, data,
3 % Template driven development n/a > 85% increasing re-usability
& code
by 85%
Standardize onboarding of new Time to instantiate a new MLOps Accelerate ML adoption
4 40 days < 1 hours
teams and ML use cases infrastructure & ML projects across all business areas
Execute the ML solutions without Your data is safe in your
5 Ensure high security standards n/a No internet
internet access in a private cloud private cloud

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation Expected Outcomes
STANDARDIZE OPERATIONS AND INFRASTRUCTURE FOR YOUR DATA SCIENCE

MLOps Expected
Business Goal Technical Metric Before MLOps Business Value
Outcomes
Time to value up to Improve Speed-to-Value
1 Be more efficient in delivery < 3 months
(from idea to production) 12 months by 4x
Time to productionize existing ML use Reduce FTE overhead
2 Simplify route-to-live 3-6 months < 2 weeks
cases in average 8x
Focus on innovation
Standardize infrastructure, data,
3 % Template driven development n/a > 85% increasing re-usability
& code
by 85%
Standardize onboarding of new Time to instantiate a new MLOps Accelerate ML adoption
4 40 days < 1 hours
teams and ML use cases infrastructure & ML projects across all business areas
Execute the ML solutions without Your data is safe in your
5 Ensure high security standards n/a No internet
internet access in a private cloud private cloud

Reduce platform, people and operation costs

Customer references building MLOps foundation and business benefits:

• NatWest: https://aws.amazon.com/solutions/case-studies/natwest-group-case-study
• BP: https://aws.amazon.com/solutions/case-studies/bp-machine-learning-case-study

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
How to Mature on MLOps?

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Maturity Model
Models in Production

Time to Proof
of Concept

Initial

• Standardize The world of Notebooks

experimentation
environments & processes
• Access to resources
• Track experiments

MLOps Maturity

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Maturity Model

Code

Model
Models in Production

deployment
time
l
Mode
Time to Proof rating Repeatable
pe
of Concept O

• Standardize source code

Initial repositories
• Automate model build,
and deployment tasks
• Automate governed access
• Standardize to resources
experimentation • Centralize management of
environments & processes models The world of automation and ML
• Access to resources pipelines
• Track experiments

MLOps Maturity

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Maturity Model
Development

ML defect
rate

Model
Models in Production

deployment
time g Model
ratin Reliable Test
Ope
Time to Proof Repeatable
of Concept • Introduce automatic
testing, monitoring,
and lineage tracking
• Standardize source code
• Standardize CI/CD
Initial repositories
practices and multi-
• Automate model build, Production
account deployment
and deployment tasks
• Automate governed access
• Standardize to resources
experimentation • Centralize management of
environments & processes models
• Access to resources
• Track experiments

MLOps Maturity

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Maturity Model
ML lifecycle
time
ML defect
rate

Model
del
Models in Production

deployment Mo Scalable
erating
time Op Reliable

Time to Proof Repeatable • ML lifecycle is reproducible

of Concept • Introduce automatic through standardized
testing, monitoring, templates shared across teams
and lineage tracking • Standardize infrastructure and
• Standardize source code new team onboarding
• Standardize CI/CD
Initial repositories
practices and multi-
• Automate model build,
account deployment
and deployment tasks
• Automate governed access
• Standardize to resources
experimentation • Centralize management of
environments & processes models
• Access to resources
• Track experiments Multiple teams

MLOps Maturity

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Maturity Model

del
Models in Production

Mo Scalable
erating
Op Reliable
Templatize and
Repeatable Productionize Multiple ML
Introduce Testing,
Solutions by Multiple
Monitoring, and Multi-
Teams
account Deployment
Standardize Code
Initial Repositories & ML Solution
Deployment

Establish the
Experimentation
Environment

MLOps Maturity

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps People & Processes

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Key Personas and Roles
Platform Team Business
Secure Cloud/Data/ML Platform Viz Dashboards, ML Adoption, & ROI

Advance Analytics Team Data Science Team

Data Lake Experimentation & MLOps
MLOps Engineer/Admin Business Stakeholder
Standardize CI/CD, user/service role, Product Owners
model consumption, testing and Define business problem, business KPIs, and
deployment methodology make business decisions
Data Engineer Data Scientist
Prepare & Ingest data Create the best ML models
building ETL pipelines to solve business problems

Security Business Stakeholder

Assess data, user, and service Data & ML Consumers
access creating policies and Consumers of ML results from other BUs,
guardrails driving business decision making
Data Owners ML Engineer
Manage data sharing Collaborate with DS to Risk & Compliance
and provide access productionize ML
Approve & Review Models

Architects/ SysOps Engineer

Standardize account infrastructure,
connectivity, user roles Auditors/Risk & Compliance
implementation Review models, data sources,
code artifacts
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation People & Processes
SEPARATION OF CONCERNS IS KEY FOR SUCCESS

ML Governance Product Owner Model Auditors &

Lead Data Scientists Approvers Compliance
MLOps System Centralized model and code artifact
Engineers Administrators storage/versioning/auditing

ML Production
Environment
Security

Data Scientists Data Scientists ML Engineers

Platform Model
Experimentation Model Build Model Test
Administration Deployment
Prove that ML can solve a business
Automate model build/training Automate model testing and Serving and monitoring the model
problem, i.e. PoC
providing scaled data guardrails testing

Provision infrastructure Ingest data Data

Provide user access Prepare, combine and catalogue Data Data Engineers Data Owners Business Stakeholders
Provide data access Visualize data ML Consumers

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation Roadmap
Technology

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation People & Processes
INITIAL PHASE

ML Governance Product Owner Model Auditors &

Lead Data Scientists Approvers Compliance
MLOps System Centralized model and code artifact
Engineers Administrators storage/versioning/auditing

ML Production
Environment
Security

Data Scientists Data Scientists ML Engineers

Platform Model
Experimentation Model Build Model Test
Administration Deployment
Prove that ML can solve a business
Automate model build/training Automate model testing and Serving and monitoring the model
problem, i.e. PoC
providing scaled data guardrails testing

Provision infrastructure Ingest data Data

Provide user access Prepare, combine and catalogue Data Data Engineers Data Owners Business Stakeholders
Provide data access Visualize data ML Consumers

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Initial Phase
ML EXPERIMENTATION ON AWS USING AMAZON SAGEMAKER STUDIO NOTEBOOKS

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker Studio
ML EXPERIMENTATION ON AWS USING AMAZON SAGEMAKER STUDIO NOTEBOOKS

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker Experiments
ORGANIZE, TRACK, AND COMPARE MACHINE LEARNING EXPERIMENTS

Tracking at scale
Track parameters and metrics across experiments
and users

Custom organization
Organize experiments by teams, goals, and
hypotheses

Visualization
Easily visualize experiments and compare

Metrics and logging

Log custom metrics using the Python SDK and APIs

Fast iteration
Quickly go back and forth, and maintain high-
quality

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation People & Processes
REPEATABLE PHASE

ML Governance Product Owner

Lead Data Scientists
MLOps System Centralized model and code artifact
Engineers Administrators storage/versioning/auditing

ML Production
Environment
Security

Data Scientists Data Scientists ML Engineers

Platform Model
Experimentation Model Build Model Test
Administration Deployment
Prove that ML can solve a business
Automate model build/training Automate model testing and Serving and monitoring the model
problem, i.e. PoC
providing scaled data guardrails testing

Provision infrastructure Ingest data Data

Provide user access Prepare, combine and catalogue Data Data Engineers Data Owners Business Stakeholders
Provide data access Visualize data ML Consumers

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
ML Solution Lifecycle Automation
CREATING AUTOMATED WORKFLOWS ON AWS – ML PIPELINES

Move from the world of Notebooks… …to the world of automation

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker Processing
PREPROCESS DATA & OFFLOAD DATA SCIENTISTS ON MANAGING CONTAINERS

Container

Code
Inputs
Outputs

Supported pre-built processors: +

Custom Containers
Framework Processors

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker Pipelines
AUTOMATED WORKFLOWS TO PRE-PROCESS, TRAIN, EVALUATION, & REGISTER MODELS

Step Types
• Processing
• Training
• Tuning
• CreateModel
• RegisterModel
• Transform
• Condition
• Callback
• Lambda Function
• ClarifyCheck
• QualityCheck
• Amazon EMR
• Fail

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker Model Registry
STORE, VERSION, AND TRIGGER MODEL PROMOTION

Create model groups in Benchmark and observe the model Track the activities, metrics, and
your model registry versions in your model group and settings per model version
promote versions of the model by
changing their status

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker Projects
EXAMPLE REPOSITORIES & CI/CD PIPELINES PER SAGEMAKER PROJECT

Feature Branch
Feature and Development branches can be used
Feature Branch for the development of new features/versions
of the Sagemaker Pipelines
Model Registry
Development Branch
Main Model Group
Main branches merge requests deploy the
Model Build Repository Sagemaker Pipelines which will be used to v1 v2 v3
(re-)train the production ready models
AWS CodeCommit ml-training-
Production Ready
Models

Main Branch pipeline

1. Validate 2. Code Level

3. Build 4. Build 5. Execute
repo structure Test e.g. unit
Custom Sagemaker Sagemaker
meet static code
Model Build CI/CD Containers Pipelines Pipeline
requirement analysis
AWS CodePipeline

Main Branch
Main branches contains endpoint
Model Registry configuration
Changes
configuration and testing code
Main Model Group New model
version event committed to
Model Deploy Repository in registry the main branch
Pre-Production Env Production Env
AWS CodeCommit v1 v2
Production Ready
v3 testing

Models Amazon
EventBridge 2. Model 5. Deploy
1. Manual 4. Manual
Endpoint 3. Testing to
Approval Approval
Deployment Production

AWS Service Catalog Model Deploy CI/CD

Product AWS CodePipeline Endpoint Endpoint

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Initial Phase
ML EXPERIMENTATION ON AWS USING AMAZON SAGEMAKER STUDIO NOTEBOOKS

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Repeatable Phase
FROM RESEARCH NOTEBOOKS TO ML PIPELINES & AUTOMATION

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Foundation People & Processes
RELIABLE PHASE

ML Governance Product Owner Model Auditors &

Lead Data Scientists Approvers Compliance
MLOps System Centralized model and code artifact
Engineers Administrators storage/versioning/auditing

ML Production
Environment
Security

Data Scientists Data Scientists ML Engineers

Platform Model
Experimentation Model Build Model Test
Administration Deployment
Prove that ML can solve a business
Automate model build/training Automate model testing and Serving and monitoring the model
problem, i.e. PoC
providing scaled data guardrails testing

Provision infrastructure Ingest data Data

Provide user access Prepare, combine and catalogue Data Data Engineers Data Owners Business Stakeholders
Provide data access Visualize data ML Consumers

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Repeatable Phase
FROM RESEARCH NOTEBOOKS TO ML PIPELINES & AUTOMATION

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Reliable Phase 1/2
UNDERSTAND THE NEEDS OF MLOPS & INTRODUCE AUTOMATED TESTING

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Reliable Phase 1/2
UNDERSTAND THE NEEDS OF MLOPS & INTRODUCE AUTOMATED TESTING

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Reliable Phase 2/2
BEST PRACTICE ON MLOPS, ROBUST AND SECURE MODELS

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker Model Monitor
AUTOMATICALLY DETECT DATA AND MODEL QUALITY DRIFTS

Inference Results
Input data

predictions
input features input
(single row) features
Historical data Inference Results
Bucket
Prediction
ut s
n p
i ure target Probability (0-100%)
t
Historical Data fea Amazon SageMaker
Model
Amazon
Bucket
Training Job SageMaker
Endpoint
Data Drift
Detection

Training Amazon
Amazon features Model Metadata SageMaker Monitor
SageMaker Baseline Statistics Bucket Scheduled Job
Monitor (.json)
Model Drift
Detection

Training Amazon
Amazon features + target SageMaker Monitor
SageMaker Baseline Amazon Scheduled Job
Monitor Statistics (.json) SageMaker
Model Registry

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
ML Governance – Model Dashboard
UNIFIED MONITORING ACROSS ALL YOUR MODELS IN PRODUCTION

SageMaker Model SageMaker

SageMaker Models Monitor Clarify
(Real time & Batch)

Business Product
Owners
SageMaker
Model Dashboard
Gain visibility into ML models
through a single pane
Lead Data Scientist

Model Card SageMaker

(Risk Rating) Lineage
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved. Tracking
MLOps Foundation People & Processes
SCALABLE PHASE

ML Governance Product Owner Model Auditors &

Lead Data Scientists Approvers Compliance
MLOps System Centralized model and code artifact
Engineers Administrators storage/versioning/auditing

ML Production
Environment
Security

Data Scientists Data Scientists ML Engineers

Platform Model
Experimentation Model Build Model Test
Administration Deployment
Prove that ML can solve a business
Automate model build/training Automate model testing and Serving and monitoring the model
problem, i.e. PoC
providing scaled data guardrails testing

Provision infrastructure Ingest data Data

Provide user access Prepare, combine and catalogue Data Data Engineers Data Owners Business Stakeholders
Provide data access Visualize data ML Consumers

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Reliable Phase
BEST PRACTICE ON MLOPS, ROBUST AND SECURE MODELS

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Scalable Phase
MULTIPLE TEAMS AND ML USE CASES ADOPT MLOPS

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Scalable Phase
MULTIPLE TEAMS AND ML USE CASES ADOPT MLOPS

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOPs Foundation Technology
MULTIPLE TEAMS AND ML USE CASES ADOPT MLOPS

Large architecture but data scientists need to be

enabled only on ….

Amazon SageMaker

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Demo

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
The Data Reply UK MLOps Journey

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
REPLY in A Nutshell
REVENUES AND PEOPLE
Over 10,000 people worldwide
€1.89 bn revenue in 2022
Founded in 1996, Reply is a company
specialized in System integration, Digital
Services and Consulting with a focus on
solutions design and implementation.

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Reply and AWS Partnership
A PASSION FOR TECHNOLOGY

Reply is among the companies with the most extensive number of competencies, including
in Data and Analytics and Machine Learning, Service Delivery Validations, Certifications
and programmes participation.

Data Reply is a Launch partner of the AWS MLOps Competency in 2021

13 16
AWS Service
9 700+
AWS AWS
Delivery AWS
Competencies Certifications
Programmes
Validations

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data Reply and AWS MLOPs Partnership

OUTCOMES
Data Reply is an AWS trusted partner in
MLOps, through AWS MLOps partner
MLOps Platform Implementation
enablement program
9-12 Months

We developed an MLOps accelerator 3 Months

using native AWS services and
ML Route to Live
successfully leveraged with a number of
3 Months
our customers
<2 Weeks

Onboarding Data Science Teams

2 Months
<2 Weeks

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Accelerator
MLOps Accelerator is a framework which aims to industrialise and operationalise ML lifecycle on AWS using best
practices and Amazon SageMaker as the key principles.

MODEL TESTING MODEL

EXPERIMENTATION MODEL BUILD PLATFORM DATA
& DEPLOYMENT GOVERNANCE

Amazon SageMaker: Amazon SageMaker: Amazon SageMaker: Amazon SageMaker: Amazon SageMaker: Amazon SageMaker:
• Studio • Processing • Endpoints • Model Cards • Custom Project • Data Wrangler
• Experiments • Pipelines • Shadow Testing • Dashboards Templates • Feature Store
• Model Registry • Model Monitor AWS Service Catalog
• Project

AWS CodePipeline GitLab GitHub CDK Jenkins Python

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Why Choose Our MLOps Accelerator

It allows customer Based on AWS Modular and reusable Adheres to AWS's Validated & Trusted
to implement the native services, but across different use best practices (AWS by AWS MLOps
MLOps capabilities can be integrated cases and WAF ML Lens) Solutions Architects
up to 50% faster with your existing environments
cloud platform

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Case Study:
ML Modernisation Challenge:
•
•
Gap in capabilities to productionise ML use cases
Manual input to deploy models
& MLOps •
•
Slow speed of development
Frustration for internal teams, lack of collaboration
• Lack of cloud operations for ML platform

Solution:
• Transitioned ML development from on-prem to Amazon
SageMaker
• The MLOps journey began with the MLOps Assessment
followed by MLOps solution implementation
• AWS Sagemaker services used to automates the build, test,
and deploy models in production
Client:
Indicia Worldwide, a global marketing agency.
Results:
Goal: • Successfully productionised 15 use cases on the cloud, with
To standardise and scale ML capabilities on AWS several more in the pipeline.
improving productivity, repeatability and time to • Reduced ‘time-to-market’ by 60-70%
market with acceptable ROI • Data scientists can focus on clients , not infrastructure
• Happier, more productive staff

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Customer Quotes
The platform that Data Reply built for us has massively reduced our time to value. What
used to take months, we can now do in weeks. It’s not just about the models. It’s about
testing, deploying, and scaling, as well as integrating them into existing systems.

Graham Lannigan
Head of Data Platform at Indicia Worldwide

Data Reply UK is helping us with streamlining and standardising our end-to-end ML processes.
With the new MLOPs Framework leveraging Amazon CDK and Amazon SageMaker services,
we are set to productionise 10 ML models in the next 6 months , scaling to tens of models
next year. This will help with optimising the recommendation system on the TUI's web app
and the website leading to an increased customer engagement and a seamless customer
experience!

Stefan Grossman
Customer Analytics Lead at TUI

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Our MLOps Engagement Approach
FOCUSING ON BUSINESS VALUE
50
Increased number
of ML use-cases
Speed to Value & Return
& Secure on Investment
Scalable
Environment

10

ASSESS ACCELERATE ELEVATE

4 weeks 6 – 12 weeks Project Dependable

Discovery & Multiple use cases leveraging MLOPs Accelerator Scale further through the business
MLOPs Solution Roadmap

Initial Repeatable Reliable Scalable

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Discovery Workshop Offer
A HALF DAY WORKSHOP AT ZERO COST TO THE CUSTOMER

Workshop overview

Why attend?
Outcomes
A customer is looking to accelerate and
scale their ML initiatives but has limited • Key ML development challenges and pain points identified
capability, skills and resources • Understanding MLOps value
• Initial use case(s) for MLOps is identified
• Stakeholders identified for the MLOps project mobilisation
Who should attend?
Head of Data Science/ML, Head of Data
Platform, Engineering, DevOps

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Thought Leadership & Additional Resources
Data Reply is a Launch Partners Of The AWS MLOPs
category Of The ML Competency. These clients have trusted Data Reply with MLOps

Our MLOps APN Blog Post – Leveraging

Whitepaper MLOps on AWS

Scan for more information

about the Data Reply
Our offering is featured on Our video Case Study with MLOps Capability
AWS Marketplace Indicia Worldwide is on Assessment
APN TV

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Resources

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MLOps Resources

MLOps on AWS by Amazon SageMaker

https://aws.amazon.com/sagemaker/mlops

Amazon SageMaker documentation

https://docs.aws.amazon.com/sagemaker/index.html

MLOps foundation roadmap for enterprises

https://aws.amazon.com/blogs/machine-learning/mlops-foundation-roadmap-for-enterprises-with-amazon-sagemaker

Secure multi-account MLOPs platform source code examples

https://github.com/aws-samples/sagemaker-custom-project-templates/tree/main/mlops-multi-account-cdk

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Thank you! Please complete the session
survey in the mobile app

Dr Sokratis Kartakis Ajish Palakadan

/in/sokratis-kartakis /in/ajish-palakadan
@sokratis.kartakis

© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.