Talk Ideas

[ian-whitestone]

• [ ] Building data pipelines with Dask and Glom
• [x] Building a home security system with Pi & Slack!
• [ ] Toronto Housing trends and analysis
• [ ] Trying to make money in daily fantasy sports
• [ ] Toronto apartment hunting with Python
• [ ] Why division is the best data science algorithm

Toronto Python Meetups:

• Serverless python with zappa [september 24, 2019]
  • nice features: packaging up python dependencies, adding a rule to keep package warm, cleaning up deployment files, setting up API gateway, getting pre-compiled lambda packages
  • show before (bash scripts)/after (with zappa)
  • demo
  • annoyances/challenges: https://github.com/ian-whitestone/domi/issues/58
• individual pieces of toronto apartment finder
  • web scraping 101
  • using glom to process results
  • using sqlalchemy to store results
  • posting results in slack

[ian-whitestone]

The Full Stack Data Scientist

• Talk about what data science means at Shopify
• How data scientists can work with product teams and have the most impact
• How organizations can set up their data scientists for success
• How data scientists can be involved at all stages of the product journey
• Develop each argument with real work examples
  • Highlight products I have worked on at each stage of the product roadmap and how data science has played a part in each
  • Talk about how our infrastructure supports scaling data science across the enterprise

Key Messages

• Data science projects rarely start with: "Build me a model that predicts X".
  • They always start with a specific business problem, for which data science expertise may or may not be a possible solution
• By embedding data scientists within product teams, you can better identify & surface opportunities for data science projects
  • This comes at the humbling of certain data scientists, who need to accept the need for more important work, like monitoring product performance or building foundational data infrastructure
  • Without an embedded data scientist, the only conversations between data science + product teams will come when a product manager has the idea that they need a model to be built for X...
• You're solving the wrong problem
  • People always ask for the solution, but often times there is a better approach to the underlying problem. countless example of this: i want to list partner id & partner name on this report. cool, why? i want to be able to quickly attribute high churn rates to particularly partners (if they come up)
  • TODO: find a better example of an adhoc product ask
  • HSC classification problem: This is not a HS code classification problem. zoom out, it's a duties calculation problem. When you change your perspective, a host of new solutions come in..
• Don't wait for someone to ask you to do something
  • Example: what country to use for currency recommendations (product team was set on using credit card bin (partly cause that's what amazon did)
• Prioritize quick data quality checks, early on
  • examples: shop_alternate_locales hard deletes...
  • examples: json parsing of monorail duties events

Practical Examples

Multi-currency

• Project scoping:
  • FX risk analysis
  • Shop segmentation
  • KPIs and targets
• Data infrastructure work:
  • Eventing/clickstream instrumentation
  • Data modelling
• Launch + product monitoring
  • dashboards
  • slack announcements
  • tracking against targets, etc.
• Post launch analysis
  • causal inference for impact analysis
• Data Products
  • cross border defaults
  • cross border recommendations via homecards
• Comparing currency risk --> bootstrapping tests

Translations

• Project scoping:
  • potential languages
• Data infrastructure work:
  • monorail instrumentation
  • Data modelling: admin language dimension
• A/B testing
• Launch + product monitoring
  • dashboards
  • slack announcements
  • tracking against targets, etc.
• Post launch analysis
  • regression for retention rates
• Data Products
  • smartling turnaround time predictions

Data Science Infrastructure

• Data acquisition

  • eventing/clickstream

• Running python jobs

  • ETL jobs
  • ML jobs