Building Refer.me

Presented at Codeaholics HK on Wednesday 11th July 2012.

An account of building Refer.me as a project using;
Ruby, Rails, PostgreSQL, ElasticSearch, Memcache, Heroku, Amazon S3, Bing Image Search, Google Geocoding API, and a lot more.

Transcript

1. Building refer.me 11th July 2012 - Codeaholics.hk @MatthewRudy
2. Good bye Armen!
3. The Site
5. The Features
6. refer.me • Ask - ask for referrals • Refer - suggest things to your friends • Save - collect your favourite things • Explore - find interesting things around you
7. Ask • Ask a Question • Send it to your friends • Listen to their Suggestions
8. Refer • Find something you love • Tell your friends about it • Share it on Facebook
9. Save • Collect your favourite things • Share them with your friends • Share them with the world
10. Explore • Surf content • Dig down by location and category • Do some clever algorithmic stuff
11. The Challenges
12. Avoiding Duplicates - Challenge • We just want one definitive version of “Carrots” • But maybe there is also a film called “Carrots” • Or a book • Or a restaurant
13. Avoiding Duplicates - Solution • Crowd source it • Always offer an existing item first • Adding a new one is a last resort • Bubble the best content to the top
14. High Quality Photos - Challenge • The design is very visual • We need good photos to pull it off • But we’ll lose users if its complicated
15. High Quality Photos - Solution • Make it really easy to add an image • Search for the item’s name • Bing Image Search • Paste a URL and grab the images
16. Good Content - Challenge • We want detailed, specific data • Users just want to click and go
17. Good Content - Solution • Take control of content after its added • Incentivise Users to be Curators
18. Everything has Location - Challenge • We want to do Geosearch • But some Items don’t have a location • Doing lat-long calculations everywhere isn’t ideal
19. Everything has Location - Solution • Define a discrete set of “Localities” • Geocode everything; • Users, Items, Questions, News Feeds, IPs • Assign objects to their nearest locality
20. Technology
21. The Back End • Ruby 1.9.x • Rails 3.2.x • PostgreSQL 9.1.x
22. Core Libraries • Devise - for authentication • Carrierwave - for uploads • Tire - for search • Delayed Job - for background workers • Rspec - for testing • Rails Admin - for CRUD
23. Deployment • Heroku Cedar • Amazon S3 • Memcachier • Bonsai Elasticsearch • New Relic
24. The Front End • HAML • SASS • Coffeescript • Twitter Bootstrap • Backbone.js
25. Geocoding • Google Geocoder • Maxmind GeoIP
26. Other APIs • Bing Image Search API • Facebook Open Graph • Mailchimp’s MandrillApp
27. Thanks to The Team
28. Efren, Victor, David, Siu Shan, and Andrew
29. Check out refer.me!