The Uber Challenge -- SF Movies

This is a Full Stack implementation of the web coding challenge found: here

The finished application can be found here: http://54.241.0.47/

Who Am I?

• LinkedIN
• github
• assembla

Technologies Used

• Java 1.7
• Spring Framework (MVC, Dependency Injection, etc)
• ElasticSearch
• JQuery
• jsonp
• sass
• jsp
• Google Maps API

Technologies NOT Used But Probably Should Have

• Maven
• Backbone.js

Why Did I Build It This Way?

I have extensive experience in everything used here with the exception of ElasticSearch. I've been using SOLR for the last few years, and I kept hearing about ElasticSearch. This project was the perfect time to break it out and see for myself.

The Java/Spring/JQuery stack is something I've been using for many years and I've built a number of personal projects on this stack. I made it clear in the Java packaging (org.gardella vs. phor.uber) what is new and what is boilerplate from years past. The javascript/css/jsp is almost entirely new.

Tradeoffs

I underestimated the quality of the data. In order to make it useful, I had to spend some time getting it into ElasticSearch in a usable way. Location data had to be harvested without proper addresses. Google rescued me there. I would like to have spend more time thinking how to make the tool more fun. It's not the best UI to inspire discovery. I did try a few things, but I knew from experience how much time that can take to get right. I had to keep it simple.

Deployment

• Tomcat 7
• AWS -- EC2 -- Single Spot Instance

Testing

• Junit test harness can be run from ant: $ant clean test
• Intern JS test can be run from /test/run-local.js

What Is Missing?

• N-Gram support for better autocomplete
• Scheduled Job Process to periodically re-run the indexer to pull new data from the SF-GOV website
• Mobile/Tablet presentation logic
• Integration with IMDB and Google Image Search, because...why not?
• Promises. Because Promises are cool.