Fernando Melo

Created a set of microservices that allowed a Brazilian fashion retailer to sell all the products from its more than 200 stores and warehouses online throughout the country.

I started as the back-end developer, and then as the project evolved, I was responsible for designing the architecture and business rules. With my comprehensive knowledge of the project, I became the natural lead developer, guiding three other developers in adding new features and optimizing existing ones.

The main API would split the shopping cart into different parts if necessary to use the best-located origins available, considering the customer's location, stores with more items available, the stores' capacity to fulfill the order, delivery price, etc.

It was created to be multi-country and multi-brand. It is used in Brazil and Uruguay but built to be flexible for adoption in other countries. It is also used in Brazil by other brands of the same company, but it was created to eventually be packaged and sold in a SaaS-like format to any other eCommerce business.

Built an API that allows customers to search all the products covered by a price comparison tool and view their prices in many different eCommerce businesses and marketplaces of the Brazilian market.

It has an indexing service running job to index the product catalog and price information into ElasticSearch, reading from a SQL Server database. It also has a querying service to get the data from ElasticSearch based on the words typed in the search box and filters applied.

I designed mappings, word analyzers, and search queries for ElasticSearch and was also one of the back-end developers of the Java API.

Created an Android app to receive payments for a drugstore chain using a Smart POS terminal. The app was integrated with the store's sale system to know the payments that needed to be made and to return the status of those payments.

It was implemented to use both the Wi-Fi network and the mobile network. That way, the store could still sell and receive payments even if the cable internet connection had a problem. The store's sale system can communicate with the POS through Wi-Fi because the local network is still working and only missing the internet connection. The Smart POS can communicate through the mobile network with the payment processing system.

Created the engine for a prize draw system using the results from the official Brazilian lottery (Loteria Federal) as a reference. The campaign was intended to stimulate sales promoters. They received coupons based on their sales. There were many rounds for the prize drawing and rules to establish how the prizes would be drawn. There was nothing random that allowed the picture to be audited.

Helped build a Magento ECommerce application customized to selling OOH (Out of Home) ad space on elevators/lifts. I was initially involved in the project thanks to my Elasticsearch experience. I customized the query used by a Magento Elasticsearch plugin to allow users to search available spots based on location (filter closest, filter by neighborhood, city, state, etc.). I had to create the front end to display this customized search result, including showing the results using a map. It allowed the customer to buy a slot in a specific building or many slots across a big region.

I also created an auxiliary application using PhalconPHP that allowed customers to upload their ad content. It was integrated with the Magento application to allow only registered users in the Magento database to list purchases.

Created an API and search engine used by a health insurance company to review hospital bills in search of improperly billed items. It allowed the reviewers to easily search multiple PDF files to check the regulations and norms enforced on a specific date.

These files were read and indexed to Elasticsearch. With the adoption of the API, every review and conclusion began to be recorded to an SQL server and indexed to Elasticsearch. This allowed reviewers to see that a similar medical bill had been reviewed before. The reviewer will then save time by checking if it's the same scenario, and they can use the same conclusion for their bill.