Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Big data and Machine learning APIs
Search
Sam Bessalah
December 03, 2014
Technology
4
260
Big data and Machine learning APIs
Sam Bessalah
December 03, 2014
Tweet
Share
More Decks by Sam Bessalah
See All by Sam Bessalah
Streaming Platforms
samklr
0
350
Intro to Parquet (June 2015)
samklr
0
290
High Performance RPC with Finagle
samklr
1
180
Dotscale 2015 Lightning - Distributed Systems Research
samklr
1
800
Datageeks_27-05.pdf
samklr
0
53
Scalable Machine Learning
samklr
2
240
mesos.devoxx.2014
samklr
2
260
Algebird : Abstract Algebra for Big Data Analytics.
samklr
9
2.9k
Algebra for analytics
samklr
1
290
Other Decks in Technology
See All in Technology
JSConf JPのwebsiteをGatsbyからNext.jsに移行した話 - Next.jsの多言語静的サイトと課題
leko
2
180
クラウドとリアルの融合により、製造業はどう変わるのか?〜クラスメソッドの製造業への取組と共に〜
hamadakoji
0
360
生成AI時代のPythonセキュリティとガバナンス
abenben
0
110
Wasmの気になる最新情報
askua
0
180
AWS UG Grantでグローバル20名に選出されてre:Inventに行く話と、マルチクラウドセキュリティの教科書を執筆した話 / The Story of Being Selected for the AWS UG Grant to Attending re:Invent, and Writing a Multi-Cloud Security Textbook
yuj1osm
1
130
Kubernetes self-healing of your workload
hwchiu
0
380
[2025年10月版] Databricks Data + AI Boot Camp
databricksjapan
1
240
アウトプットから始めるOSSコントリビューション 〜eslint-plugin-vueの場合〜 #vuefes
bengo4com
3
420
「魔法少女まどか☆マギカ Magia Exedra」のIPのキャラクターを描くための3Dルック開発
gree_tech
PRO
0
150
Linux カーネルが支えるコンテナの仕組み / LF Japan Community Days 2025 Osaka
tenforward
1
120
MCP ✖️ Apps SDKを触ってみた
hisuzuya
0
300
Dylib Hijacking on macOS: Dead or Alive?
patrickwardle
0
450
Featured
See All Featured
Building Better People: How to give real-time feedback that sticks.
wjessup
369
20k
The Myth of the Modular Monolith - Day 2 Keynote - Rails World 2024
eileencodes
26
3.1k
Speed Design
sergeychernyshev
32
1.2k
Done Done
chrislema
185
16k
Building Flexible Design Systems
yeseniaperezcruz
329
39k
StorybookのUI Testing Handbookを読んだ
zakiyama
31
6.2k
Performance Is Good for Brains [We Love Speed 2024]
tammyeverts
12
1.2k
I Don’t Have Time: Getting Over the Fear to Launch Your Podcast
jcasabona
34
2.5k
Visualization
eitanlees
149
16k
Documentation Writing (for coders)
carmenintech
75
5.1k
Producing Creativity
orderedlist
PRO
347
40k
Fight the Zombie Pattern Library - RWD Summit 2016
marcelosomers
234
17k
Transcript
Big Data and Machine Learning APIs
Sam Bessalah @samklr Software Engineer, Freelance Data Engineering, Distributed systems,
Machine Learning Paris Data Geek Meetup @DataParis me :
None
None
None
Big Data Legends ….
Big Data Legends … Web logs Sensors Other Data source
.. . . .
A Big Data Legend … Web logs Sensors Other Data
sources .. . . .
A Big Data Legend … Web logs Sensors Other Data
sources .. . . .
A Big Data Legend … Web logs Sensors Other Data
sources .. . . .
A Big Data Legend … Web logs Sensors Other Data
sources .. . . .
A Big Data Legend … Web logs Sensors Other Data
sources .. . . . Data Driven Decisions Smart Applications
BUT ….
- Building big data infrastructures is no easy task. -
Leveraging data for decision making requires a mix of multiples skills : . System Engineering . Distributed computing . Statistics . Machine Learning
Solutions …. - Build Data platforms as a service. -
Build robust and consistent APIs to bring big data to the masses. - Leverages fluent APIs for fast data science
None
Big Data is not just about throwing data to Hadoop.
It’s also about data pipelines
Data Sources
Data Sources
Data Sources - High Throughput distributed mssaging platform - Publish
Subscribe Model - Modelled as a distributed replicated log - Persists messages to disk - Categorizes messages into Topics - Allows message retention for long specified amount of time - Allows stream replay in case of failure
Data Sources Machine Learning High Latency Batch Apps Real Time
Processing
How do you build an API around that?
None
/ingest REST API
/ingest
/ingest /query /trainModel /process
Things to be careful with - Multitenancy (Yarn, Mesos, Docker…)
- Job Scheduling - Security - Serialisation : ProtoBuf, Thrift, Avro - Storage Format : Optimize queries with columnar storage. - Compression : LZO, Snappy
Making sense of data …
None
What is Machine Learning?
http://dilbert.com/strips/comic/2013-02-02
None
https://speakerdeck.com/nivdul/lightning-fast-machine-learning-with-spark-1
Machine Learning workflow
Machine Learning workflow Text, Images, etc
Machine Learning workflow Text, Images, etc Feature Extraction
Machine Learning workflow Text, Images, etc Feature Extraction Learning algorithm
Training
Machine Learning workflow Text, Images, etc Feature Extraction Learning algorithm
Training Predictive Model
Machine Learning workflow Text, Images, etc Feature Extraction Learning algorithm
Training Predictive Model New Data Feature Vector Prediction
Machine Learning workflow Text, Images, etc Feature Extraction Learning algorithm
Predictive Model New Data Prediction
Machine Learning workflow Text, Images, etc Feature Extraction Learning algorithm
Predictive Model New Data Prediction BLACK BOX
Machine Learning Libraries and Frameworks
scikit-learn.org
Text, Images, etc Feature Extraction Predictive Model New Data Prediction
X = vect.fit_transform(input) clf.fit(X,y) X_new = vect.fit_transform(input) y_new= clf.predict(X_new)
http://arxiv.org/abs/1309.0238
From library to web APIs
Machine Learning workflow Text, Images, etc Feature Extraction Learning algorithm
Predictive Model New Data Prediction BLACK BOX
Machine Learning workflow Text, Images, etc Transformed Data Application Prediction
Predictive API
Predictive Web APIs
Some examples
Challenges of Predictive APIs
http://www.r-bloggers.com/data-science-toolbox-survey-results-surprise-r-and-python-win/
Modeling and Prediction are just a small part of the
process
- Data locality and data gravity - Support the full
workflow - Verticalization of platforms - Scalability - Collaboration and interoperability - Black boxing of implementations
Explore machine learning for APIs orchestration. Talk to Ori @OriPekelman
Next Frontier ? Or actual reality ?
None
http://speakerdeck.com/samklr