Daniel Pereira

I built a micro-services architecture along with a team of eight other Software Architects. This architecture is message-driven, using RabbitMQ as the broker mechanism. It has services such as user management, authentication, authorization, customization, configuration management, scheduling, translation, cache, and much more. It also includes features like Single Sign-On, and it has support for both REST and SOAP services.

We can deploy the platform in both cloud and on-premises environments, and it doesn't have vendor lock-in with any cloud provider. Developers deploy new versions using the pipeline provided by the CI Server, using Docker and Kubernetes.

This platform generated hundreds of new services inside the company, and it is allowing all legacy systems to be available in the cloud.

I made an enterprise social network, distributed in the SaaS model. People would create publications and other people interacted by liking it or adding a comment.

The social network's primary focus was to share relevant news across the company. Different from an intranet portal, it gives employees the power to interact with the publication. It also allowed them to create the content themselves, which increased engagement by a wide margin.

Early stats collected from the service revealed that engagement increased by more than 200% compared to legacy solutions.

In Brazil energy is expensive on weekdays, since most industries only work from Monday to Friday, thus making the demand very high. Since the energy provider doesn't have a high demand on weekends, they incentivize companies by charging less money on Saturdays and Sundays.

One of our clients sells rice, and they have a very automated environment that could run with very few people working, which means they could benefit from this incentive. To make it happen, they need their silos to be empty by Friday afternoon, so the rice could be processed and store during the weekend. Since verifying the silos were empty was a manual process, it was often not completed by Friday afternoon. On these occasions, they would not be able to produce during the weekend, which did not allow them to take advantage of the lower energy rates.

I made a solution that verified the silo level in real-time and would warn if the silo needed to be emptied by Friday afternoon. We had sensors to measure the percentage of the silo that was filled, and we had a dashboard to keep track of this information. This project saved the client a considerable amount of money since they could produce on the weekends, reducing their energy costs.

We had a system that was migrated from Delphi to Java. This migration was made by a converter that kept the original structure from Delphi and used JNI to make the native calls. Both languages have very distinct architectures, which led to inferior performance in the Java version. I worked to reduce both memory usage and increase performance on this system.

This project helped me understand how the JVM works under the hood for subjects like Garbage Collection. It also gave me a lot of experience with tools that analyze the program memory and performance, such as Memory Analyzer and JVisualVM. After the project was finished, the memory and performance were vastly improved, allowing the Java version to replace the Delphi version in production.

This project enabled the company to remove Delphi from their stack, which was a goal they had from a long time.