| Oct 2021 - Present |
ML/AI Engineer at Jumio Corporation, Italy / Austria
Research and Development for Bank Document Verification from Photos |
| June 2020 - Oct 2021 |
Post-doctoral Researcher at Multimedia Group Unitn, Italy
Research in Graph Neural Networks for Computer Vision and Unsupervised Learning |
| May 2019 - Nov 2019 |
Deep Learning Intern at IBM Research, Zurich
Graph Neural Networks for record linkage and data modelling, Demo video. |
| June 2018 - Aug 2018 |
Machine Learning Intern at IBM Benelux, Amsterdam
Active learning and adaptive text classification algorithms. |
| Oct 2017 - Dec 2017 |
Deep Learning Intern at Pi School, Rome
Ransomware detection through analysis of user logs and RNN models, Demo video. |
| Jan 2016 – May 2016 |
Data Fusion intern at Purdue University, USA
Data fusion from noisy sources of information for object lifespan and ground truth inference. |
| Feb 2015 - Sep 2015 |
Python Software Engineer at iTechArt Group, Belarus
Dvelopment endpoints for RESTful web services. |
| Aug 2014 - Feb 2015 |
Embedded Software Engineer at Promwad Electronics, Belarus
Embedded software for one board Linux computers and microcontrollers. |
Hello! Since October 2021 I work as a Machine Learning engineer at Jumio Corporation, where I work on Document Classification and Data Extraction from Personally Identifiable (PII) documents, such as Bank Statements.
I was a Post-doctoral researcher in Multimedia Group at the University of Trento where my work spans over Graph Neural Networks for Computer Vision, Domain Adaptation, and Self-supervised Learning, in particular, when little or no supervision is available.
I hold a PhD (Cum laude) in Computer Science from the University of Trento in 2020, where I worked on cost-aware active learning, crowdsourcing and hybrid crowd-machine learning algorithms for low-risk classification decisions.
In my work, I use a variety of Machine Learning and Deep Learning approaches, such as CNN, ResNet, Deep Cluster, Byol, GAN, PointNet, RNN, Transformers, BERT, and Graph Neural Networks. For obtaining high-quality training data from (possibly) noisy sources, I implement state-of-the-art crowdsourcing and data fusion algorithms.