37signals is the company behind the well-known Basecamp and, more recently, Hey e-mail. It has been at the forefront of software development and remote work culture for decades and has chosen to be profitable since the 90s. Responsible for the conception of Ruby on Rails and other open-source tools, 37signals shared a lot with the community and continues to be one of the organizations that provoke changes in the industry.

Projects:

• Scale Prometheus monitoring in the cloud by running Thanos and its components
• Revamp Prometheus on-prem monitoring by recreating the whole setup

Intricately is a data company based in the US that provides go-to-market technical information for anyone who sells products to tech companies. After a few weeks in Intricately, it got sold to HG Insights, a company in the same market.

Projects:

• Build terraform modules for AWS Backup and Vault monitoring
• Evaluate possible solutions for migrating an on-VM MongoDB to a managed solution
• Tackle security tasks reported by a third-party organization

QuintoAndar is an online real estate services company similar to Airbnb but for the long term. It is the second largest startup in LatAm, just behind Nubank. It has around 300 micro-services and more than 700 Kubernetes nodes in total. It has a big SRE team (approximately 70), and I was part of the observability-focused team.

Projects:

• Maintain all Prometheus instances used as the primary monitoring system. Each instance ran on top of Kubernetes and had more than 4TB of data on disk.
• Install and configure Thanos to scale Prometheus monitoring through a lot of different Kubernetes clusters and improve how efficiently we could store monitoring data for the long-term

Pier is the first Brazilian Insurtech, trying to solve the problem related to the bureaucracy and lack of high granularity data-driven decision making in the insurance world. At Pier, we needed to start structuring the DevOps/SRE culture from scratch, bringing some useful practices and tools to the table like post mortems, incident response, key engineering metrics like MTTD, MTTR, success rate and latency.

Projects:

• Start monitoring apps using Newrelic APM
• Build the logging stack using ElasticSearch, FluentD and Kibana
• Define the logging pattern
• Organize PostgreSQL replicas, users and overall maintenance
• Secure AWS account and IAM related configuration

As ContaAzul’s platform is cloud based, all our customers access the same production environment and the number of customers is always growing we need to understand and apply some important concepts and practices in order to improve our overall service reliability. Besides being responsible for our cloud infrastructure, the SRE team is always looking into keeping the DevOps culture warm.

Some of this practices are:

• Post Mortems, to keep learning from failure
• Automation, to scale and keep things consistent
• SLIs, SLOs and SLAs, to build a contract between our product and development teams seeking balance
• Release Engineering, to guarantee that changes to the production environment have a controlled impact
• Monitoring, to enable true understanding of what is happening in production

We currently own a self-managed Kubernetes cluster deployed to AWS responsible for running different kinds of application containers (microservices, jobs etc) and our data is mainly saved on PostgreSQL instances and S3

Tech: Kubernetes, Docker, Prometheus, AWS, Grafana, Opentracing, Jaeger, Java

Team: Billing & Payments

Our goal is to allow any customer using ContaAzul to exchange money through our platform. At ContaAzul, we have a lot of specific strategies related to payments. As no available platform met our requirements we opted to build and maintain our own payments infrastructure from the UIs that allow our customers to choose and pay using their preferred payment methods to the back-end that securely calculates things like subscriptions, recharge and discounts.

We have some core values respected and followed by everyone in the team:

• Quality
• Reliability
• Testability
• Software engineering principles

Our back-end microservices are mainly in Java (SprintBoot) and our front-end uses AngularJS with Material Design.