Saturday, Mar 16th, 2019
● For today
○ Build an application with google assistant and Cloud functions
○ Build a social wall completely Serverless with Firebase and GCP
○ Serverless machine learning at DYNO
Speakers
How to get
free labs?
hint: using code
How to get free labs?
Step 1: Join Qwiklabs
How to get free labs?
Step 3: Access this link
https://go.qwiklabs.com/next-19-extended
How to get free labs?
Step 2: Buy it…. just kidding!
How to get free labs?
Step 4: Get started
How to get free labs?
Step 5: Enter the code
How to get free labs?
1t-hanoi-1234
Hands-on Labs
Build a Youtube Entertainment
App
Case Study
Building a serverless social wall with
Firebase and GCP
Lim Shang Yi
Software Developer @ IdealHub
Google Developer Expert, Firebase
@buggybyrd
@jarrodek
Why build a social wall?
Why build a social wall?
● Use for events
● Increase social media presence
● Fun activity before the event starts
● Customize it any way you like
Features of the Social Wall
✅ Pull public feeds from Twitter and Instagram, every 2 minutes
✅ New feeds will always appear in front of the list
✅ Moderate posts automatically/manually on monderator screen before
showing
✅ Ability to adjust number of lanes on the fly, depending on screen resolution.
✅ Adjust header and subtitle in real time.
✅ Ability to configure hashtag on the fly
✅ Ability to configure how fast/slow new posts appear on screen
✅ Ability to configure how many posts can be queued before showing.
How would you build it?
How would you build it?
● Layouts
● Styles
● Animations
● Pull public feeds from Twitter and
Instagram every 2 minutes
● Moderate posts before posting
● Update hashtags, header on the fly
● Update client configurations
● Hosting client app
Front end Back end
How would you build it?
● Layouts
● Styles
● Animations
● Pull Twitter/
Instagram every 2 minutes
● Manage data
● Websocket server
● ...
Front end Back endAPI
● Store
data
How would you build it?
● Layouts
● Styles
● Animations
● Pull Twitter/
Instagram
every 2 minutes
● Manage data
● Websocket
server
● ...
Front end Back end
● REST API
● Websocke
ts
API DB
Application Server
● Store
data
How would you build it?
● Layouts
● Styles
● Animations
● Pull Twitter/
Instagram
every 2 minutes
● Manage data
● Websocket
server.
Front end
Back end
● REST API
● Websocke
ts
API DB
Building with serverless
Application Server
● Store
data
Typical architecture
● Layouts
● Styles
● Animations
● Pull Twitter/
Instagram
every 2 minutes
● Manage data
● Websocket
server.
Front end
Back end
● REST API
● Websocke
ts
API DB
Application Server
● Store
data
Serverless architecture with Firebase and GCP
● Layouts
● Styles
● Animations
● Pull Twitter/
Instagram
every 2 minutes
● Manage data
● Websocket
server.
Front end
Back end
● REST API
● Websocke
ts
API DB
Firebase
● Store data
● Real-time
connection
Serverless architecture with Firebase and GCP
● Layouts
● Styles
● Animations
● Pull Twitter/
Instagram
every 2
minutes
Firebase
Hosting
Cloud
Functions
Cloud
Firestore
● Trigger
Cloud
Functions
Cloud
Scheduler
How it works
How it works
triggers every
2 minutes
Cloud
Functions
#cat
Pull tweets
Pull instagram
posts
Sanitize, and
adds social post
to Firestore
How it works
Client
Pull Real-time
social posts
from Firestore
Firebase SDK
Administrative
Panel
Firebase SDK
Authenticated
Moderate
posts/delete
social post
Key takeaways
● Developer effort reduced for key functionalities
● Reliable and scalable performance
● Secure database
● Ability to easily extend services
Thank you
http://bit.ly/eoa-client
http://bit.ly/eoa-functions
Serverless Machine Learning with GCP
Dao Tuan Vu - vu@dyno.vn
CTO DYNO - EWAY JSC
About Me
- CTO DYNO - Eway JSC
- Experience on backend, database, big data and
machine learning
About Eway
- DYNO.vn: Big data services
- AdFlex.asia: Top 1 CPA network in Viet Nam
- MasOffer.net: Top 1 CPS network in Viet Nam
- iHR: Big data for recruiting
- Kalapa: Fintech AI
- eDoctor.io: Platform for healthcare
Machine Learning Flow
1. Understanding of business problem
2. Problem formalization
3. Data collecting
4. Data preprocessing
5. Modeling
6. Way to evaluate model in real life
7. Way to deploy model
Machine Learning Flow
1. Understanding of business problem
2. Problem formalization
3. Data collecting
4. Data preprocessing
5. Modeling
6. Way to evaluate model in real life
7. Way to deploy model
Cloud Vision
Cloud Vision
Translate API
Other services
Machine Learning Flow
1. Understanding of business problem
2. Problem formalization
3. Data collecting
4. Data preprocessing
5. Modeling
6. Way to evaluate model in real life
7. Way to deploy model
If you want to customize model
AutoML
https://towardsdatascience.com/google-cloud-automl-vision-for-medical-image-classification-76dfbf12a77e
AutoML
Develop a Machine Learning Model From Scratch?
Developing Model
Developing Model
Develop a Machine Learning Model From Scratch?
AI platform (ML engine)
Machine Learning Flow
1. Understanding of business problem
2. Problem formalization
3. Data collecting
4. Data preprocessing
5. Modeling
6. Way to evaluate model in real life
7. Way to deploy model
Data processing
Exploratory Data Analysis
1. Cleansing: Checking for problems with the collected data, such as missing
data or measurement error, data type of columns, etc.)
2. Defining questions: Identifying relationships between variables that are
particularly interesting or unexpected.
Exploratory Data Analysis
More data?
Datalab + Bigquery
Datalab + Bigquery
Datalab + Bigquery
Scaling with Dataflow
Machine Learning Flow
1. Understanding of business problem
2. Problem formalization
3. Data collecting
4. Data preprocessing
5. Modeling
6. Way to evaluate model in real life
7. Way to deploy model
Dataprep
1. Assessing your data quality
2. Resolving any issues uncovered
3. Validating big data
Data Profiling
Data Profiling
Data Profiling
Resolve issues
Resolve issues
Extract data
And more ...
Validate big data
Machine Learning Flow
Demo
https://eway.vn/tuyen-dung
jobs@eway.vn
THANK YOU

Google Cloud: Next'19 Extended Hanoi