Bernardo Sandino Azevedo de Amorim

An electronic trading system to manage the operations of Brazil's largest quant hedge fund.

I was the sole developer of this project—coded entirely in Python—which allowed traders to both test and deploy their trading models through a simple Python interface.

A marketing tool to allow marketing staff to create and compare the performances of different variations of the same ad template, which could be made both in the "video" or "playable" formats.

With this tool, we transformed the performance comparison of different ads from a 3-squad manual process, that took weeks to complete, into a completely automated and parallelizable pipeline.

The back end was implemented in Go with a Python AWS Lambda-hosted template-rendering tool and a React-based UI.

An educational platform to host prep courses for the Vestibular exam, the entrance exam used by Brazilian universities. The platform comprised a PWA app—both mobile and desktop compatible—and an internal area, and it received 10,000+ unique daily users.

I managed the tech staff in the implementation of the product while also doing a substantial amount of coding myself. We also chose the stack, which included Node.js and PostgreSQL for the back end (exposing a GraphQL API), React for the front end, and AWS-backed Kubernetes for the infrastructure.

A service to handle profile management and authentication for an educational platform's customers and employees, with a total of over 100,000 unique daily visitors.

I managed a tech team of four that was responsible for implementing and maintaining the accounts ecosystem which was comprised of an authorization service, profile management service, and web app.

We implemented the back end in Go and PostgreSQL with the front end in Vue.js, and everything was hosted via AWS ECS.

An Android app to allow diabetes patients to track their daily activities and glucose measurements. Besides being able to input exercises and meals and chatting with health experts, the patients could seamlessly input glucose level data by integrating their app with their glucose meter's Bluetooth server.

All patient data was managed by a Node.js and PostgreSQL server and made available to the experts on a React-backed internal app. A GCP-backed Kubernetes cluster managed the entire infrastructure.

I led and managed the development of the Android app, the server, and the infrastructure. I also hired a developer to do the iOS version.

The tricky part of this project was the Bluetooth integration: each glucose meter vendor exposed its own Bluetooth interface—which was often undocumented—making the team have to use Wireshark to reverse-engineer the integration pipelines.

Integration between the company's CRM and interactive voice response (IVR) systems made it possible to automatically render the CRM profile of each incoming caller to the company's support line onto the attendant's screen.

I was the sole designer and developer of the solution.

There were two main challenges with the implementation:

1. I had no prior Asterisk experience. I bypassed my lack of knowledge by implementing a simple Python script and installing it into the Asterisk server that sniffed the call logs and sent the appropriate events to our CRM service.

2. Since the CRM was web-based and we wanted to send server-side events to the attendants, we had to expose WebSockets. WebSockets would be spread out through the many instances of the CRM service, which introduced a routing challenge. We eventually solved this by using Redis Pub/Sub channels.

An app that allows teams to create and analyze sprint retrospectives. I helped with the development of both back and front-end. The server side was implemented using typescript-powered NestJS, with MongoDB as the persistence layer, while the front-end was developed using Angular