Overview

Disappointed that I missed an opportunity to attend my first AWS re:Invent conference (I had a ticket but changed jobs before I could go!), I decided to take to YouTube to catch up on all of the announcements and deep dive breakout sessions. When searching, I found myself overwhelemd and frustrated with irrelevent search results and outdated content. I just wanted a single spot to browse re:Invent specific videos, and have the ability to quickly and easily find what I was looking for.

The website reinventvideos.com was built to solve exactly that problem.

Here's the infrastructure in a nutshell:

Backend

The backend is powered by the serverless framework. It builds a few components:

• A Lambda function that is triggered by an Amazon CloudWatch event. When run, it queries the YouTube API to get latest statistics about re:Invent sessions. It combines that data with data stored in a DynamoDB table which holds information about the available podcasts for the sessions. When complete, it aggregates the data and updates Algolia search indexes.
• A second Lambda function that is triggered when a new file is put into the S3 /podcasts/ directory. This function will add a new item into DynamoDB with the session, year, and object key information. This data is used by the previous Lambda function and added to the Algolia indexes.

Frontend

The frontend is powered by a single page app written in Vue.js. It leverages the Vue InstantSearch plugin to interface with Algolia. The UI is built with latest version of Bootstrap.

Serving the application is Amazon CloudFront with Amazon S3 as the origin. DNS is hosted by Amazon Route 53.

Future Enhancements

Time permitting, I'd like to extend functionality of the website to include:

• Allow users to create accounts and save / favorite videos for later viewing
• Provide powerpoint decks alongside the sessions
• Make certain parts of sessions discoverable, e.g. jump to Announcement X from a keynote