We use Chef for our configuration management and our service discovery.

We rely on SNS for Push delivery to Android & iOS.

We're in the process of moving to a SOA (service-oriented architecture), and each of our services run on a cluster of EC2 boxes behind an ELB. Our services are built primarily with Go and Ruby, depending on the requirements.

Our entire infrastructure is located in Amazon VPC.

We use Packer to build our Base AMIs.

We use Vagrant for our development machines and to run our Chef integration tests.

We use Elasticsearch to power our Relevance algorithms.

We use PostgreSQL to store most of our data.

We wrote our own image processing, resizing, and snapshotting service in Go to allow our clients to send photos and GIFs to each other. Files are stored in S3, resized on the fly using OpenCV, and then cached in GroupCache before being served to clients.

Go allows it all to be quite fast and efficient, and entirely non-blocking on uploads!

We use Rails to power most of our APIs and internal services.

We use Ruby for our APIs, coupled with Rails. We also use it for scripting and other open source applications. Check out https://github.com/cotap/stacker and let us know what you think!

All of our web apps are built using BackboneJS. Keeps our code clean and organized, and keeps the page refreshing to a minimum!

John, thanks for the update. At the time we working to become HIPAA compliant, AWS was not yet including RDS in their BAA. Even now, they do not include RDS when running the PostgreSQL storage engine—which is the one we use. It's important to regularly check with AWS though, as they frequently update the list of services covered by their BAA as they update services and release new ones.

UPDATE: As of Nov 29, 2016, RDS running the PostgreSQL storage engine is now HIPAA compliant: https://aws.amazon.com/about-aws/whats-new/2016/11/amazon-aurora-and-amazon-rds-for-postgresql-are-now-hipaa-eligible-services.