Hello, I am Albert Calvo

Industrial Machine Learning: Safe and Robust AI

Fraud, Anomalies and Unusual Behaviors Detection

Tech Transfer: From Science to Market

Trustworthy AI: Usable and Explainable AI

I am PhD Student in Computing at Universitat Politecnica de Catalunya. My Phd "Explainability in Data Science-oriented Industrial Projects", focus to build interpretable proxies, allowing corporations and industries to take advantage of actual state-of-the-art algorithms without sacrificing transparency and confidence.


Let’s connect! For further collaborations and projects you can reach me at albert@cs.upc.edu

Some of my research topics are the following:

Fraud, Anomalies and Unusual Behaviors Detection

Fraud is one of the principal sources of economic losses in industries and organizations and to identify and mitigate and mitigate the threats is a priority. Artificial Intelligence has become a swiss knife allowing to analyse large amounts of data in unprecedented ways. During the last years, I have been researching the application of Machine Learning for Fraud Detection in the Energies sector, specifically the detection of Non-Technical Losses. Recently, I made a switch into the digital world researching cyberthreats.


Artificial Intelligence has become a key-player in almost all economic sectors from Agriculture, Energies or the Financial sector uses Artificial Intelligence to optimize and automate processes to a more productive and competitive economies. If the decision making is done without supervision it can carry undesired outcomes causing bias. Explainability is the science of comprehending what the algorithm has learnt providing an user-friendly explanation to the Stakeholder allowing to work with Black-Box models (High Accuracy) without sacrificing interpretability.


Work Experience

AI Research Engineer (2020 - )

Distributed Artificial Inteligence (DAI) - i2CAT

Research Assistant (2015 - 2019)

Process and Data Science UPC Group (PADS-UPC)

  • Fraud Detection in Energy Consumption In this project are applied different state-of-the-art Machine Learning techniques for optimizing the waste of energy produced by fraud, irregularities and structural network problems. During this project are applied several techniques to face problems such as quality of the data, imbalanced datasets or scalability of algorithms.
  • Activity recognition using low-price sensors The objective of this project was to identify different activities in domestic environments for patients with dementia.

Junior Web Developer (2014-2015)

Facultat de MatemĂ tiques i EstadĂ­stica UPC

Deployment of an intranet for student administration. I work mostly with web and database Technologies such as Code Igniter, a PHP Framework, JQuery and SQL.


Phd in Computing (2019 - Now)

Universitat Politècnica de Catalunya (UPC)

  • Explainability in Data Science-oriented Industrial Projects

Master in Innovation and Research (2016 - 2018)

Barcelona School of Informatics (FIB-UPC) • Barcelona, Lausanne

  • Data Science specialization
  • Exchange semester at École polytechnique fĂ©dĂ©rale de Lausanne (EPFL)
  • Attendance to the e-health Eurocampus Summer School 2017 at Institut National Universitaire Champollion.
  • Dissertation: Classification of Virtual Patent Marking web pages using ML techniques . Efficient identification of web pages that contain Virtual Patent Marking information under time and resources constraints. Virtual Patent Marking allows owners of products publish product-patent information online.

Master’s Degree Program in Communications and Information Systems Management (2016 - 2017)

Universidad Politécnica de Madrid (UPM) • Madrid

  • Dissertation: Design and implementation of Business Intelligence Systems

Degree in Informatics Engineering (2011-2016)

Barcelona School of Informatics (FIB-UPC) • Barcelona

  • Information Technologies specialization
  • Dissertation: Deployment of Spark at MareNostrum III Supercomputer This project evaluates several data-intensive workloads on MareNostrum III; a Supercomputer with a peak performance of 11.15 Petaflops. During this project are analyzed different ML algorithms using Spark.