I work as a front-end developer in the APIA team, mainly responsible for the invoice module in the B2B application of Mercado Eletrônico.Currently, I work with Vue.js 2 and implement features, tests and fixes for several micro-services on the platform.I also participate in planning and decisions related to the front-end of applications.

On Nelogica, I developed several features and fixes for products aimed at customers in the financial market, more specifically, those who operate with stocks and cryptocurrencies. Here, I have the opportunity to work with the development of several products at the same time, namely, Vector, Profit, Phoenix and BlackArrow. In my day to day, I mainly use Vue.js, TypeScript, Javascript, WebSockets, HTML, CSS, SCSS, and other tools.

I joined Buzzmonitor initially as an intern, where I stayed for about 1 year and five months, and after that, I was hired (jr developer) and stayed for 3 months. Here, I had the opportunity to work with a platform focused on companies that seek to have insights and monitoring of their social networks in a simpler and easier way. I was able to work with the API of several social networks (facebook, instagram, Reclame Aqui, twitter, etc) and I was able to gain a lot of knowledge in the front-end development area. I worked mainly with AngularJS (Version 1), SCSS, CoffeeScript and Ruby on Rails. I had the opportunity, still as an intern, to develop several interesting projects, such as the new home of the application, in which I dedicated the period of 3 months developing, because it was a page that displays the content of all the sessions of the application, and in that I could have a good experience with project architecture.

In this project, I worked mainly testing and researching new optimization techniques for Deep Learning models (project focused mainly on classification/detection in images), with the objective of keeping them usable (with equal or lower accuracy without losing much quality) but with a lower computational requirement. I studied some optimization techniques, transfer learning, etc. The objective of this project was to make these highly accurate models able to be used in more limited environments, such as small hardware (e.g raspberry pi).I also had the opportunity to work with full stack web development in creating a service that was used to visualize and test the models obtained. I developed a page that displayed the model's detections, as well as real-time metrics, given an RTSP stream or even a video.In this project, I mainly used the technologies: React + Redux, Flask, Tensorflow, PyTorch, Javascript, HTML, CSS.