2. Innovate in 2019
Artificially Intelligent, Constantly
Connected and Intuitive at Hyper-Scale
Olivier Klein
Head of Emerging Technologies Asia-Pacific
6. ARTIFICIAL INTELLIGENCE AND
DATA-DRIVEN DECISIONS
U R G E N T
X-Ray analysis
GAN rendered
dental implants
Blood sugar measurements
through breath
MRI scans analysis
Deep Learning
assisted ultra-sound
19. AWS CloudTrail – “Cloud” usage logging & automatic compliance
You are making
API calls...
On a growing set of
services around the
world…
AWS CloudTrail is
continuously
recording API
calls…
And delivering
log files to you
U S E R A C T I O N T I M E
T I M
C R E A T E
D
1 : 3 0 P M
S U E
D E L E T E
D
2 : 4 0 P M
K A T
C R E A T E
D
3 : 3 0 P M
C L I
C L O U D F O R M A T I O N
E L A S T I C B E A N S T A L K
C O N S O L E
Config
24. Containers / Serverless
Web Standards
Frameworks
APIs / Microservices
CLIENT
PORTABILITY
BACKEND
PORTABILITY
INFRA
PORTABILITY
SDKs
Cross-Platform
Native Apps
25. $ amplify init
$ amplify add hosting
$ amplify publish
A comprehensive library for building
sophisticated serverless cloud-powered
apps
Works well with JS frontend libraries like
React, Angular, Vue.Js or Ionic
Web apps, Hybrid Mobile Apps, Static
Websites and Progressive Web Apps (PWA)
JS Library for Cloud Apps Development
28. Thinking about data as an asset, not a cost
Stop
throwing
data away
Make it
available to
more users
Arm users
with more
data processing
technologies
29. Glacier
Deep Archive
Lowest cost storage available in the
cloud…even lower than on-premises
tape
No tape to manage
Designed for 99.999999999% durability
Recover data in hours vs. days/weeks
$0.00099/GB/month
Less than 1/4th the cost of Glacier
30. AWS Marketplace
Amazon Redshift
Data warehousing
Amazon EMR
Hadoop + Spark
Athena
Interactive analytics
Kinesis Analytics
Real-time
Amazon Elasticsearch service
Operational AnalyticsRDS
MySQL, PostgreSQL, MariaDB,
Oracle, SQL Server
Aurora
MySQL, PostgreSQL
Amazon
QuickSight
Amazon
SageMaker
DynamoDB
Key value, Document
ElastiCache
Redis, Memcached
Neptune
Graph
Timestream
Time Series
QLDB
Ledger Database
S3/Amazon Glacier
AWS Glue
ETL & Data Catalog
Lake Formation
Data Lakes
Database Migration Service | Snowball | Snowmobile | Kinesis Data Firehose | Kinesis Data Streams | Data Pipeline | Direct Connect
Managed
Blockchain
Blockchain
Templates
Amazon
Comprehend
Amazon
Rekognition
Amazon
Lex
Amazon
Transcribe
AWS DeepLens 250+
solutions
730+ Database
solutions
600+ Analytics
solutions
25+ Blockchain
solutions
20+ Data lake
solutions
30+ solutions
RDS on VMWare
Broadest and deepest data and analytics portfolio
BUSINESS INTELLIGENCE & MACHINE LEARNING
DATABASES ANALYTICS
DATA MOVEMENT
DATA LAKE
BLOCKCHAIN
32. *Since re:Invent 2017
Improvements to
speed
Compiled code cache
Support for lateral column alias reference
Query processing improvements
Query planning
Commit processing enhancements
Short query acceleration
Query rewrites that pushdown selective joins into a subquery Resource management for memory-intensive queries
Queries that refer to stable functions with constant expressions
Improvements for the COPY operation when ingesting data from Parquet and ORC formats
Performance improvement for queries that refer to stable functions over constant expressions
Hash join performance
Joins involving large numbers of NULL values in a join key column Late materialization
Queries operating over CHAR and VARCHAR columns
Expressions on the partition columns of external tables
2x the number of tables in a cluster
DC2 nodes
Single-row inserts
Queries with intermediate subquery results that can be distributed Faster string manipulationResult caching
Complex EXCEPT subqueries
COPY operation when ingestion data from Parquet and ORC formats
Concurrency scaling
33. 87% of Amazon Redshift
customers don’t have
significant wait times
Daily
cluster
queue
time per
day
Remaining 13% have
bursts of activity
averaging 10 minutes at
a time
34. Amazon Redshift
Concurrency Scaling
Consistently fast performance
with thousands of concurrent queries
0
5000
10000
15000
5 40 80 120 130 180
Amazon Redshift
Amazon Redshift with Concurrency Scaling
35.
36. Robotics is undergoing fundamental
change in collaboration, autonomous
mobility, and increasing intelligence
• Logistics
• Construction
• Retail
• Hospitality
• Healthcare
Robots are being put to work every
day across many industries
• Agriculture
• Energy Management
• Oil and Gas
• Facilities Management
• Household chores
41. • Map villages in Africa to deliver right
amount of vaccines
• Provide first responders with
information during national disasters
• PSMA Australia - 200TB imagery - 7.6
million km2 - 20 million buildings
Cloud-based platform to derive insights
and convert unstructured satellite
imagery into meaningful insights using
ML
Generates 80TB/day of imagery data
44. E ASILY ADD IN T EL L IGENCE TO APPL ICATIONS
N O MACH IN E L E AR N ING SKIL L S R EQUIRED
MACHIN E L E ARN ING FOR E V E RY DEVELOPER AN D DATA SCIE N TIST
BU IL D, T RAIN AN D D E PLOY ML FAST
CH OICE AN D FL E XIBIL IT Y W IT H BROADEST FR AME W ORK SUPPORT
FAST EST AN D LOW EST -COST COMPUT E OPT IONS
AI Services
ML Services
ML Frameworks
+ Infrastructure
45. R E K O G N I T I O N
I M A G E
R E K O G N I T I O N
V I D E O
T E X T R A C T
VISION
P O L L Y T R A N S C R I B E
SPEECH
T R A N S L A T E C O M P R E H E N D
LANGUAGE
L E X
CHATBOTS
F O R E C A S T
FORECASTING
P E R S O N A L I Z E
RECOMMENDATIONS
AI Services
ML Services Amazon
SageMaker
Ground Truth Notebooks Algorithms + Marketplace Reinforcement Learning Training Optimization Deployment Hosting
ML Frameworks
+ Infrastructure EC2 P3
& P3dn
EC2
C5
FPGAs Greengrass
Elastic
inference
FR A ME WO R KS I NT E R FA CES I N F R A S T R U C T U R E
Inferentia
EC2
G4
46. AWS is framework agnostic
Run them fully managed Or run them yourself
49. AWS Deep Learning
Containers
Quickly set up deep learning environments with
optimized, pre-packaged container images
Runs on Kubernetes, Amazon ECS, and Amazon
EKS
Packaged and configured by experts
Includes latest version of popular frameworks:
(PyTorch
coming soon)
Available in the AWS Marketplace and
Amazon EC2 Container Registry
51. S T A T I O N A R Y C A R
S T A T I O N A R Y C A R M O V I N G C A RP E D E S T R I A N
S T O P S I G N
D I R E C T I O N
S I G N
S T O P
S I G N
T R A F F I C
S I G N A L S
T R A F F I C
S I G N A L S
S T A T I O N A R Y V E G E T A T I O N
S T A T I O N A R Y R O A D
S T A T I O N A R Y B U I L D I N G
M O V I N G
C A R
P A V E M E N T
Successful models require high-quality training data
53. Amazon SageMaker Ground Truth: How does it work?
Human
annotations
Data in S3
Automatic
annotations
Training
data
Simple, pre-built workflows
Mechanical Turk
Third-party vendors
Your own employees
Active
Learning
model
>80% confidence
<80% confidence
54. How do you teach machine learning
models to make decisions when there is
no training data?
56. Learn by
interacting with
the real world
Model problem
as a simulation
environment
Trial and error
Observe results
Optimise learning
strategy to maximise
long-term reward
57. Vehicle routing
Objective Fulfill customer orders
STATE Current location, distance from homes …
ACTION Accept, pick up, and deliver order
REWARD Positive when we deliver on time
Negative when we fail to deliver on time
Applicable in many domains and industries
58. Amazon
SageMaker RL
New machine learning capabilities in
Amazon SageMaker to build, train and
deploy with reinforcement learning
Fully managed RL
algorithms
TensorFlow, MXNet,
Intel Coach, and Ray RL
2D and 3D simulation
environments via OpenGym
Simulate with Sumerian
and AWS RoboMaker
Example notebooks
and tutorials
59. Build reinforcement learning model
DeepRacer League Races at AWS Summits
Winners of each DRL Race and top scorers
compete in Championship Cup at re:Invent 2019
Virtual tournaments through the year
AWS DeepRacer League
World’s first global autonomous racing league
60. and technology is changing the world
The world of technology is changing