<aside> 🚧 This section is still work-in-progress. Soon we will provide more insight about how we work in the product engineering team in detail.
</aside>
We follow a lean and agile way, taking the best principles from SCRUM
and Kanban
. Soon we will publish more details on how we run.
Since we are working with a high-load
and low-latency
real-time
architecture, we are paying a lot of attention to high test coverage.
Developers are responsible for writing tests themselves. Integration tests are written and run continuously against testing and live environments. QA tests are written by the team continuously. Learn more here:
We have a fully integrated CI/CD
managed in Kubernetes
, all standardised and automated. This allows us to do daily deployments. To guarantee zero downtime and a high NPS
we have integrated canary deployments
via Kubernetes
which allows us to perform incremental roll-outs.
When releasing, we want to run the new and the old versions side by side to track the quality of the new release. Once we have confirmed that NPS
and average session duration are the same or better for a particular commit hash, we move over. If they are worse or we experience a spike in error messages, we fix them before continuing to roll out or even rollback.
We have an On-call rotation in place, with DevOps/senior devs being on call.
“You build it - You run it!”
We run on AWS
and consider moving parts of our high-load
architecture to own managed servers.
<aside> 🔥 Looking for a new challenge in DevOps? We are looking for a ‣ 😇
</aside>
The Wonder platform is built on the success of open-source
software. We encourage our team to give back to the community as much as possible.
Since we are driving a lot of innovation especially in the WebRTC
space we will have some interesting stuff we consider publishing in the future.