Software Engineer in Brasília.
Software developer with around 4 years of backend experience in Python, Django, PostgreSQL, and Google Cloud Platform. I build ERP modules, services consumed over REST APIs, and asynchronous integrations for the digital ecosystem of a restaurant chain.
Click a technology to see where I used it.
2024 – Atual Current
Restaurant chain
2022 – 2024
Restaurant chain
Model Context Protocol agent specialized in my work context. Performs code reviews with a custom skill based on articles about code review and AI. Graph-based documentation in Obsidian, where each node represents a module, submodule, or feature.
Pluggable logging module for an ERP. It lets you pick at runtime which modules ship logs to Google Cloud Platform, rebuilding the configuration on every change, with no system restart and no coupling to the code of the monitored modules. The point is dynamic configuration: adding or removing a module from monitoring is a configuration change, not a code change.
3 of 4 modules shipping. Configuration rebuilt without a restart.
Loyalty program service consumed by multiple applications over a REST API. It handles customer scoring and ranking, point redemption rules, and a configurable rewards engine, such as progressive discounts above a given purchase amount. Rules can be set by period, by a customer milestone date (a birthday, for instance), and by usage limits, so a new campaign never requires a code change.
Backend layer that manages the virtual wallet of a food ordering app. It keeps customer balances and financial movements, and client applications consume it over a REST API. The work is about the reliability of the service layer behind every transaction, not the interface.
Locust load testing suite that simulates concurrent access to validate the performance and scalability of a service before it goes to production. It exists to find the bottleneck and set a safe capacity limit while that is still cheap, instead of discovering the ceiling with the feature already live.
At 150 concurrent users the load is comfortable and nothing queues up.
Menu publishing service built on Redis pub/sub messaging, with Celery workers handling the asynchronous processing. The part I like most is the onboarding: the service runs in a Docker container configured purely through environment variables, so connecting a new application to the Redis cloud and having it receive its own menu takes one folder and one configuration file, with no code change.
Two applications subscribing to the channel, each one receiving only its own menu.
Software Engineering
University of Brasília (UnB)
2020 – 2026
If something here matches what you are looking for, send me an email.