Data Scientist

• Oct 2023 - Present

Senior Data Scientist

• Feb 2020 - Sep 2023

As part of the risk team: - Fraud detection system. Developed a fraud detection system using traditional techniques (Variational Autoencoder and XGBoost) and Graph Neural Networks (GNN). With the information generated by the models, orders can be automatically approved or declined. - Implemented two scoring models based on GNN: 1. Semi-supervised classification of nodes in a directed graph. This model aims to predict if a node (an order) is fraudulent or not. 2. Multi-label graph classification task using bipartite graphs. A unidirectional bipartite graph is used to model a customer. Cross-functional projects: - Recommendation system. Developed a graph-based item-item recommendation engine. - Price-demand elasticity forecasting. Developing a demand forecast model with a focus on price by using the Temporal Fusion Transformer. Show less

Postdoctoral Researcher

• Jun 2017 - May 2019

I worked on the ATLAS experiment at the Large Hadron Collider (LHC) since 2013. ATLAS is an international collaboration seeking answers to fundamental questions such as what are the fundamental forces of nature?. In this position, I: - Worked on data analysis: Precise measurement of the properties of the top-quark using combined 2015-2018 data collected with the ATLAS experiment. To process the huge amounts of experimental data (petabytes) distributed grid computing is needed as well as the development and maintenance of complex software applications. To provide an interpretation of the underlying phenomena these physics analyses use machine learning algorithms and statistical methods like maximum likelihood estimation. - Supervised graduate and undergraduate students. Helping them to execute scientific work and to conclude their project with determination. - Performed data quality monitoring tasks useful to the whole collaboration. - Imparted knowledge to bachelor and master students in lectures and seminars. Presented results to experts, colleagues and the general public. Show less

Doctoral Student

• Oct 2013 - Oct 2016

During my Ph.D., I was involved in the analysis of data collected by the ATLAS detector at the LHC. - Worked on data analysis: Search for the associated production of the Higgs boson with a pair of top quarks. Performed the signal-to-background separation using a two-stage machine learning strategy. First, a machine learning algorithm is used to reconstruct the signal-like events, then this event probability together with global event features feed a second classification machine learning in order to separate the signal events from the huge background events. This novel approach was validated and adopted by the ATLAS collaboration. Note that due to its physics importance, this analysis is one of the most challenging and competitive topics at the LHC. - Developed a new method to identify jets containing two B hadrons using a machine learning technique (boosted decision trees). The method was successfully implemented in the ATLAS framework. - Worked in migrating large-scale scientific software to cope with the data format changes of the software infrastructure needed for the ATLAS Run 2 period (2015-2018). Show less

Internship

• Feb 2013 - Apr 2013

Worked on the track reconstruction algorithm for the vertex locator of the LHCb experiment. Performed the vectorization of the algorithm as well as the development of a new algorithm with parallelism in mind to an easy adoption in many-core architectures like GPUs. Worked on the track reconstruction algorithm for the vertex locator of the LHCb experiment. Performed the vectorization of the algorithm as well as the development of a new algorithm with parallelism in mind to an easy adoption in many-core architectures like GPUs.