Keller Williams Realty Consumer
In late 2018 I embarked on a new engineering journey, this time in Real Estate. I was hired to help out building a Cloud Infrastructure platform for the Consumer facing greenfield project at KWRI, Consumer.
We were going after Zillow, a major player in the field. Our team was amazing, ranging from designers, UI/UX enthusiasts, all the way to great full-stack JavaScript developers.
I managed to squeeze myself somewhere in the middle, as a DevOps Engineer. It was quite a ride. Jumping between Node.js and SRE depending on the requirements, I finally started practicing DevOps methodologies full-time. The team needed a person to be in charge of all the infrastructural changes that started happening being in a fast-paced development environment.
We couldn’t handle any more manual changes in IAM, non-programmatic additions to our storage buckets, deployments of new microservices (starting with 20 of them), and handling of all our serverless workloads such as lambdas and managed dockers. I decided to take full responsibility of those tasks.
I also decided to fully integrate with KW teams on separate projects to reuse their infrastructural patterns. You definitely don’t want to introduce yet another methodology if there’s an existing, working one already in-place.
To deploy all infrastructure configuration we used Terraform, an laC tool from Hashicorp. For front-facing services me have decided to go with GCP’s App Engine, which we later abandoned for GKE.
All the microservices were running on GKE, fronted with LBs and GraphQL proxies using Redis Memory Cache. We also had a few Google Cloud Functions in place. Backend was MongoDB in Atlas and Elasticsearch in Elastic Cloud, but the data in that one got migrated to Atlas, so we closed it out.
For Continuous Integration and Deployment we leveraged Travis-Cl, talking to GitHub and GKE, using tight IAM service accounts. LightStep was used for performance monitoring and logging of microservices.
It was a rollercoaster ride (ended mid 2020, with the whole team delivering on the expectations), and I learned tons on new tech but also about myself. The latter, in TL;DR, being to “Keep it Simple!”. Not all new tech solutions should be taken for granted. I should still count in proven, stable ones!
I met so many nice people and made long time friends and connections on the way. That’s invaluable! Also visited Keller Williams HQ in Austin in Dec 2019. That was the icing on the cake.
Now onto next one!
AWS Solutions Architect Associate Exam
I decided to give myself some time away from projects in order to do some certifications relevant to my current state in career, and that’d be cloud technologies. I chose to go with Amazon Web Services Solutions Architect Associate exam because I’m really interested in a higher picture of the product I’m building at a particular moment in time.
Even though I had a good chunk of experience using AWS by building my SaaS offering called Edgebind I enrolled in a great course for the exam preparation on A Cloud Guru. The content there is pretty amazing. Took me about a month to prepare for the exam and to of feel confident enough with the subject.
The exam itself was pretty tough but I managed to pass it on first try.
Bought myself a beer in my favourite bar right after the exam to celebrate. :)
Shipped my first SaaS product
This one is definitely a topic of its own. I started building Edgebind purely based on minimal demand from my customers at the time. I had no clear vision on how to monetize it or distribute it properly.
My Time to Market was huge. I was taking on the major players in industry like Balena and Mender with a competitive edge based on modularity with a convention-over-configuration mindset where customers needn’t have so much expertise in Linux, Dockers and OTA software.
It was a huge task and definitely not focused enough for a single person to handle all by themselves. One day I was building embedded Linux images with Yocto, the other day it was HTML, CSS and JS. Not to speak of cloud tech on AWS such as Cognito API gateway, lambda, ECS, IAM and security groups. Plenty of work. I learned so much. But I wasn’t focused on customers and marketing as much as I needed to. Too much emotion in building the next great IoT Edge gateway got in the way.
It all starts and ends with a customer. We must follow their needs in order to build features that’ll streamline their work. You know, customer is always right. It doesn’t matter if the backend is built with Ruby on Rails or Node.js, if it’ll run on AWS or GCP. I know “performance” is a great word, but a product doesn’t start or end with it.
The 2020 goal I had in mind was having a platform to be stable enough to be rolled out to first potential customers. Check!
app.edgebind.io is live for some time now and current customers like it. Cuts their time quite a bit. Makes them more productive. So I guess not everything is lost. Who knows what 2021 holds. Let’s persevere and see what happens!
My band’s EP
Aside from work, I spent my time writing music and lyrics for my band, Kozjak. We play hard rock, alternative metal music. A tough market fit, also, heh.
Back in 2018 we started writing new material and we ended up with four songs. We went into studio early 2020 to record everything and it went pretty smooth. Took us about 12 hours to record vocals and all the instruments, which is pretty nice. After a few revisions with sound engineers we ended up with a finished product.
I’m not planning to say the 2020 word, but it shaped our release management plans for the album. We’re aiming for the end of the pandemic to start publishing the material song after song, as separate singles. This includes radio cuts of the songs, accompanying video material and teasers for social networks such as Instagram.
Seems like musicians are half businesspeople nowadays. Or at least the market expects us to be like that. Hence so many failed attempts happening?
Let’s see how it works for us in 2021!