4. 4
Applied Science at Zynga
Goal: Build Portfolio-Scale Data Products
Example Projects
• Propensity Models
• Recommendation Systems
• Player Segmentation
• Anomaly Detection
Tech Stack: AWS, Python, Databricks, CouchBase
5. 5
Why Build a Portfolio?
Showcase your experience
Practice writing
Learn new tools
Practice deployments
Avoid take-home projects
6. 6
• Get hands-on with the cloud
• Create a new data set
• Glue things together
• Stand up a service
• Create a stunning visualization
• Write a white paper
Advice for Aspiring Data Scientists
7. 7
• Get hands-on with the cloud
• Create a new data set
• Glue things together
• Stand up a service
• Create a stunning visualization
• Write a white paper
Advice for Aspiring Applied Scientists
Themes
• Ecosystems
• Scale
• Integrations
• Authoring
9. 9
Programming Ecosystems
Interview Question
• Tell me about a time that you put a model into production.
Follow-up Questions
• How did you make updates to the model?
• What team was responsible for model failures?
• Did you consider different deployment approaches?
10. 10
Programming Ecosystems
Goals
• Get hands-on with different cloud environments
• Learn tradeoffs between different platforms and tools
Projects to explore:
• Setting up a predictive model as an endpoint
• Deploying a model as a serverless function
• Building a reproducible environment
• Persisting models to cloud storage
11. 11
Getting Started with Cloud Tools
• AWS Free Tier
• GCP Credits
• Databricks
community edition
16. 16
Computing at Scale
Interview Question
• What is the biggest data set that you’ve worked with?
Follow-up Questions
• How did you distribute the problem?
• How did subtasks coordinate?
• How would you work with live data?
• How scalable is your approach?
17. 17
Computing at Scale
Goal
• Use datasets so large that you have to use distributed methods
Approaches
• Spark
• Cloud Dataflow
• Calculations in SQL
24. 24
Integration Projects
Interview Question
• Tell me about a time where you had to integrate different systems.
Follow-up Questions
• Why was this approach necessary?
• Are there other implementations that you considered?
• How would you approach this problem now?
25. 25
Integration Projects
Goal
• Build end-to-end model pipelines
Approaches
• Use portable model formats
• Leverage new language features
• Use managed services
• Write a wrapper
31. 31
Writing about Projects
Interview Question
• How do you like to advocate for your projects?
Follow-up Questions
• What is your medium of choice?
• Have your projects been extended by others?
33. 33
Writing Topics
Learning Something New
“Deploying Keras Deep Learning Models with Flask”
Deep Dive
“Why & How Words With Friends Is Adopting React Native”
Project Postmortem
“Words With Friends 2 & How We Built Lightning Round”
34. 34
White Papers
Detailed Project Overviews
- Executive summary
- Project Overview
- Details
- Appendix
Use a platform with comments
40. 40
Portfolio Themes
Ecosystems: get hands-on with cloud computing
Scale: explore distributed computing
Integrations: glue things together
Authoring: go beyond decks
41. 41
Building an Applied Science Portfolio
Ben Weber
Distinguished Data Scientist
Zynga
bweber@zynga.com
@bgweber
https://www.zynga.com/jobs/