Resume

About Me

Software Engineer, passionate about distributed systems, machine learning, and Kubernetes. Interested in helping people take their ideas into production.

MSc in Electrical and Computer Engineering at Técnico Lisbon (IST) with a major in Systems, Decision and Control, and a minor in Computer Science. Former research intern at Carnegie Mellon University (CMU). Currently working as a Software Engineer at Cortex Labs.

Experience

  • Software Engineer, Cortex Labs, Dec 2020 - Present, Remote

    Cortex is an open-source platform to simplify high-scale application deployments on Kubernetes and AWS. https://github.com/cortexlabs/cortex

    Achievements:

    • Development of Cortex’s Kubernetes operators to manage Custom Resource Definitions (CRD) for different application kinds.
    • Design and implementation of a scale-to-zero mechanism for deployed APIs.
    • Design and implementation of an HTTP reverse-proxy to handle request metrics, such as latency and number of in-flight requests.
    • Design and implementation of an AsyncAPI kind to handle requests asynchronously, by consuming events from a queue.
    • Prometheus integration with the cluster.
    • Development of end-to-end and integration tests.

    Technologies:

    • Go, Kubernetes, Istio, Docker, AWS, Kubebuilder, Prometheus, Prometheus Operator, Grafana, CircleCI
  • Senior Machine Learning Engineer, Sensity, Sep 2019 - Nov 2020, Amsterdam, The Netherlands

    Sensity (formerly Deeptrace Labs) is the antivirus for “Deepfakes”, focused on using deep learning and computer vision to detect and monitor AI-generated media.

    Achievements:

    • Design and implementation of Sensity’s event driven, asynchronous, and micro-services based API to process incoming requests for analyzing videos and images.
    • Development of a configuration based machine learning framework in pytorch for fast model iteration aimed at reproducibility, speed of development and deployment.
    • Deployment of several computer vision models into production for media analysis.

    Technologies:

    • Go, Python, Pytorch, Docker, Google Cloud Run, Google Cloud PubSub, Firebase, CircleCI, Github Actions, Kubernetes, Knative, Istio, Keda, Cert-Manager, PostgreSQL
  • Machine Learning Engineer, BrainCreators, May 2018 - Aug 2019, Amsterdam, The Netherlands

    BrainCreators is a team of artificial intelligence experts with decades of experience of machine learning and enterprise software engineering.

    Achievements:

    • Development of state of the art speech recognition models for both English and Dutch
    • Creation of an unsupervised machine learning pipeline to assist manual labeling of network anomalies for a major internet service provider
    • Development of an object detection pipeline with Faster/Mask-RCNN
    • Contributions to several open source projects, including projects from Facebook Research and Uber.

    Technologies:

    • Python, Pytorch, Docker, Flask, GitLab CI, Travis CI.
  • Machine Learning Scientist, Unbabel, Jan 2017 - Apr 2018, Lisbon, Portugal

    Unbabel is an AI start-up which provides a translation pipeline combining machine translation with a community of human post-editors to assert the quality of the translated text.

    Achievements:

    • Part of the creation of a framework agnostic codebase for training and inference of any machine learning model.
    • Implementation from scratch of a quality estimation framework in PyTorch for post-edited text. This framework is currently inserted in the company’s pipeline.
    • Deployment of a machine translation quality estimation system into Unbabel’s pipeline, to decide whether or not to post-edit a given document, impacting the company’s cost per word in almost 4%.
    • Creation of an onboarding fake-task generator for the evaluation of new editors.

    Technologies:

    • Python, Pytorch, Flask.

Education

  • Research Intern - Carnegie Mellon University, 2016, Pittsburgh, USA

    Worked on the topic of voice conversion, advised by Prof. Alan Black, and outperformed state of the art results for parallel data voice conversion with a 1/16 th fraction of the dataset of the previous existing research.

  • MSc in Electrical & Computer Engineering - TU Delft, 2014, Delft, The Netherlands

    ERASMUS Exchange Semester

  • Integrated Masters in Electrical & Computer Engineering - Instituto Superior Técnico, 2010 - 2016, Lisbon, Portugal

    Major - Systems, Decision & Control

    Minor - Computer Science

    GPA - 16/20

    Master thesis - Voice Conversion using Deep Neural Networks. Advised by: Prof. Isabel Trancoso & Prof. Nuno Fonseca. Grade: 19 / 20

Publications

  • Segment Level Voice Conversion with Recurrent Neural Networks - Varela Ramos, Miguel & W. Black, Alan & Astudillo, Ramon & Trancoso, Isabel & Fonseca, Nuno. (2017). 3414-3418. 10.21437/Interspeech.2017-1538.