(2018 - 2020) Jumpseller - Part-time (Student-worker).

• Jumpseller is an e-commerce provider where you can easily set up your online store. During this period, I made part of their software development team. While working there I coded in Ruby, using the known framework Ruby on Rails as well as in ReactJS for frontend development. Besides this, I also worked with job queues using Sidekiq, webhooks for service connection, and with caching mechanisms using Redis. In terms of databases, I worked with both MySQL and PostgreSQL databases. Overall I gained many insights on the development and maintenance of a highly used web application as well as the software development process using SCRUM. It was my first job on the field, working part-time and managing my time as a student-worker.

(2020 - 2021) Jumpseller - Full-time.

• This is the moment I turned myself into a Machine Learning Engineer, and because it was a small company, in terms of the number of employees, I also made a lot of work in Data Engineering. Initially, I was responsible for building a large-scale Big Data collection pipeline using AWS services, namely AWS Kinesis Firehose for stream-data collection which is stored in AWS S3, processed and aggregated using AWS Glue, and queried using AWS Athena. I’ve also helped implement a more robust search mechanism for the main project using ElasticSearch.

• I worked on a large-scale online machine learning project related to recommendations. I’ve worked with content-based recommendations using a NLP deep learning neural network architecture called Word2Vec, and also collaborative filtering using item-based recommendations with cosine similarity and a session-based recommender system (GRU4REC neural network). I’ve also used ElasticSearch for this task. This project’s architecture is much different than the usual applications I’ve done so far since it’s done mostly via separated Docker images stored using AWS ECR, and because it has to interconnect with the main project which has been a challenge but very fun to do. The models are trained and served using AWS Sagemaker. It includes a deploy CI/CD pipeline which deploys the project when in the master branch using Gitlab pipelines. I used Python and some of its libraries such as Numpy, Tensorflow, and Pandas. I also want to note that a component of this project was an on-demand machine learning model which can be asked for predictions in a Flask server.

(2021 - 2023) Farfetch

I’ve started working at Farfetch in September 2021, and I was a part of Farfetch’s recommendations cluster working as a machine learning engineer in Porto. I worked together with data scientists and machine learning engineers as well as product managers in a professional environment. I’ve started using other technologies such as PySpark, Azure Pipelines and Databricks for an integrated ML and data platform. Together we built products using Approximate Nearest Neighbour Search and used the Resnet and LSTM architectures to build an outfit recommendations generator. Lastly, we were able to powere this recommender with the Vespa.ai vector database.

(2023-Now) Revolut

Currently working in the fincrime department, in a fast-paced and quick-changing environemnts building models to detect fraudulent CNP transactions. Using a lot of data analysis tools, I improved my usage of pandas and plotly.express for data manipulation and data visualization. For OLAP queries I became fluent with Trino and am using SQL on a daily basis, as well as acquired a lot of Regression Trees and Gradient Boosting Mode knowledge.