Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Sign up for free
Menu
Search
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Pricing
Search
Sign in
Sign up for free
Scaling Machine Learning at Holiday Extras (Big...
Search
Sponsored
·
Ship Features Fearlessly
Turn features on and off without deploys. Used by thousands of Ruby developers.
→
Rebecca Vickery
November 13, 2019
Technology
140
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
Scaling Machine Learning at Holiday Extras (Big Data LDN 2019))
Rebecca Vickery
November 13, 2019
Transcript
40 slides
Scaling Machine Learning at Holiday Extras REBECCA VICKERY | DATA
SCIENTIST @vickdata
Travel planning is time consuming Airport parking Airport hotels Airport
lounges Travel insurance Holiday money Port products Car hire Airport transfers 582 minutes Over 46 days* Travel Planning *Facebook commissioned consumer research company GfK
Optimising consumer decision making Airport parking Airport hotels Airport lounges
Travel insurance Holiday money Port products Car hire Airport transfers Less Hassle. More Holiday Trip recommendations
Automated bidding Ad targeting Channel optimisation 1 Ad spend 2
Commercial 3 Customer Experience 4 Marketing Lots of other processes to optimise Automated pricing Allocation Revenue optimisation Automated call handling Personalised experiences Intelligent messaging Optimise send frequency
How to scale Use Cases and Buy in (Input Team
Deployment
How to scale Use Cases and Buy in (Input Team
Deployment
“Ideas are worth nothing unless executed”, Derek Sivers
Deploying machine learning is hard Scaling is even harder
Tools - Data Scientists Open source Lack Software Development expertise
Mainly Python c Flaticon
Tools - Software Engineers Different tools Lack ML/Data expertise Mainly
Javascript c Flaticon
Data science process The wrong kind of independence c Flaticon
People Small data science team Science + software experts are
rare c Flaticon
Two types of deployment
Bespoke Solutions “Ideas are worth nothing unless executed”, Derek Sivers
c Daniel Moyo
Unused Models Many models never make it to production “Ideas
are worth nothing unless executed”, Derek Sivers
Time to model deployment Model development = days to weeks
Model deployment = weeks to never! “Ideas are worth nothing unless executed”, Derek Sivers
The technology
c Flaticon init.py task.py setup.py model.py Model Package
Repeatable, Reusable Process init.py task.py setup.py model.py Model Package
Data transformations Scikit-learn pipelines + custom transformers Transformation occurs in
the model
Solution for other libraries too Add preprocess file to the
package Image taken from Google Cloud documentation
Further customisation Custom scoring Custom prediction routines
None
Faster time to production c flaticon Fully Managed service
Not Quite!
Collaborative Project
ML Proxy (bespoke ML microservice)
Model Versioning
Monitoring - Model Performance
Monitoring - AI Platform Performance
Time to model deployment Model development = days to weeks
Model deployment = hours to days “Ideas are worth nothing unless executed”, Derek Sivers
How to scale Use Cases and Buy in (Input Team
Deployment
The right kind of independence c flaticon Data Scientists have
full ownership over models
The right kind of independence c flaticon Data scientists work
closely together
The right kind of independence c flaticon But they also
work closely with other teams
Use cases and buy in c flaticon Focus on problems
to solve
Use cases and buy in c flaticon Don’t start in
the highest value area
Use cases and buy in Deploy a first version (not
the best) as fast as possible
Test and learn Photo by Alex Kondratiev on Unsplash Use
cases and buy in
Thank you @vickdata
More Decks by Rebecca Vickery
See All by Rebecca Vickery
Pair Programming with AI
rebeccavickery
1
120
Machine Learning for Everyone
rebeccavickery
0
35
Data Preparation and the Importance of How Machines Learn
rebeccavickery
0
170
Scaling_Machine_Learning_at_Holiday_Extras_-_MUC.pdf
rebeccavickery
0
1.2k
Gender Bias, Why we Need More Women in Tech
rebeccavickery
0
1.2k
The Fastest Way to Learn Data Science
rebeccavickery
0
69
Employing Google Cloud Machine Learning Engine to Develop Models in Production
rebeccavickery
0
1.3k
Other Decks in Technology
See All in Technology
Go でトランザクション、どう書いてますか?
knwoop
1
560
農家は秘伝のタレを知らない - NOBODY KNOWS THE HIDDEN SOURCE - Kaggriculture 3rd Place Solution
masspeaks
3
680
Re:ゼロから始めるシェアハウスのペットな鯖
silmin_
0
270
20261119_AgentCore
takuyay0ne
1
200
パスワードのない鍵で、 データを手元に置く
ginjihayashi
0
400
From vibe coding to reliable (agentic) development
jcabot
0
200
データから見る AI時代のエンジニア転職と採用
junya_okura
4
1.2k
コーディングできない僕にBobが舞い降りた
shumpei3
0
190
RAGの改善を速く回すための評価基盤
yosukeyoshida
6
3.5k
Well, that’s weird…
csswizardry
1
200
AWS DevOps Agentの構造から読み解く運用 AI エージェントの設計原則/aws-devops-agent-design-principles
tomoki10
0
260
目を覚ませ!僕らの認証情報が何者かに侵略されてるぞ
tkito
2
480
Featured
See All Featured
The #1 spot is gone: here's how to win anyway
tamaranovitovic
4
1.2k
Effective software design: The role of men in debugging patriarchy in IT @ Voxxed Days AMS
baasie
1
620
Save Time (by Creating Custom Rails Generators)
garrettdimon
PRO
33
5.9k
Become a Pro
speakerdeck
PRO
31
6.4k
How to optimise 3,500 product descriptions for ecommerce in one day using ChatGPT
katarinadahlin
PRO
3
3.9k
Getting science done with accelerated Python computing platforms
jacobtomlinson
2
580
[Rails World 2023 - Day 1 Closing Keynote] - The Magic of Rails
eileencodes
38
3.1k
Code Reviewing Like a Champion
maltzj
529
40k
Music & Morning Musume
bryan
48
7.5k
Navigating Team Friction
lara
192
16k
We Have a Design System, Now What?
morganepeng
55
8.4k
Building a A Zero-Code AI SEO Workflow
portentint
PRO
0
840