All of our Technical Pile at Coffee Touches Bagel. Our company is a mobile-first and mobile-only company

Probably one of the most usual questions that meeting candidates ask whenever they interview only at coffees matches Bagel is actually, “what’s their tech stack?” it is also one of the more common concerns I get around San Francisco while going to technology meetups and seminars and such. Very for your needs curious candidates and anyone else curious exactly how a top level matchmaking providers has built their particular platform, here you go, I’ll try to break it down available.

Initially, at a high-level:

The Android os app try indigenous (Java)

Our very own apple’s ios app was native (Swift + unbiased C)

Our backend program coating was Python & Scala

We incorporate Golang for some data pipeline tasks

All of our backend is usually in AWS with a little in yahoo Cloud

We strive for totally automated evaluating

Now, let’s enter some nitty-gritty, shall we?

Exactly why are we cellular just?

We have a splash page (and a couple of advertisements web sites in some places), nevertheless only access points for the app tend to be all of our cellular Android and iOS customers. We in fact launched with best a webpage in 2012, but once we launched our iOS app in 2013, our very own website traffic dropped to something such as 40percent of your total visitors. Then, once we established our very own Android os software in 2014, site traffic plummeted to below 5percent of total traffic. When this occurs, we decided it simply was actuallyn’t worth every penny to utilize the restricted manufacturing information to compliment the internet site, therefore we fallen it and switched they into just a landing page that directs browsers to either the software shop or Play shop.

I’ll leave the specifics of our own mobile architecture for another article, hopefully as written by one of our excellent mobile engineers, but at a high-level our cellular developing philosophy is actually:

The mobile apps become “dumb” people. Their primary task occurs the data delivered by the backend in an engaging, performant, steady, and beautiful fashion. We achieve this by adhering semi-strictly to RESTful web services basics.

Divorce of problems is essential for maintainability, stability, and testability. New code written should abide by a VIPER structure (for iOS) or MVVM (for Android os).

We’re on the way to automating UI evaluating on iOS using XCTest.

We’re currently automating a lot of the UI testing, and useful screening, on Android using Appium.

third party gear in the cellular area:

apple’s ios employees makes use of Bitrise for CI

Android os teams makes use of Jenkins for CI

Firebase for real time speak

Localytics for analytics

Materials (Crashlytics) for accident reporting and beta group government

Myspace for verification

As I’ve discussed earlier, we have been about completely in AWS, for much better or bad, and we are practically completely a Python store. I’ll allow the details of your backend structure for ideally the next article from 1 your exemplary backend designers, but also for here’s a few more information.

We incorporate a number of AWS solutions:

Cloudfront to give us rapid international entry to cached tools

S3 for photograph space, facts copies, and hosting static the internet sites

SQS (straightforward Queuing solution) for queuing up numerous asynchronous work

ALB (program burden Balancer) for balancing weight and routing visitors to the appropriate service

Cloudwatch for tracking and alarming

Redshift for information warehousing

ElasticCache for maintained Redis for our algorithm group

VPC (digital personal affect) to guard our very own computers from outside community and from just one another

Kinesis to flow occasions throughout our system

We also use several different databases technologies:

Redis (Monolith — master / servant set-up) w/ HAProxy for legacy in memory space databases

Redis (Cluster) for horizontally scalable in memory space databases

Postgres w/ HAProxy for deal databases

Cassandra divide over 2 datacenters

ElasticSearch for looking / blocking means information necessities

With respect to tracking:

Sentry for error reporting

Grafana for higher level dashboards and graphs

Cloudwatch for spying and worrying everything AWS relating

Pagerduty for controlling the on-call schedule and awakening you right https://datingmentor.org/spiritual-dating-sites/ up in the middle of the night time

Opscenter for monitoring Cassandra

Kibana for record aggregation

Kopf for ElasticSearch tracking

New Relic for servers / system wellness tracking

Other 3rd Party Treatments server-side:

Localytics for statistics and giving pushes

Mailchimp for giving emails

Branch for connect generation and attribution

Myspace for verification and shared pal computations

Firebase for real-time talk

Jenkins for creates, deployments, and scheduled activities

Imgix for picture caching and in-the-cloud changes