Professional Experience

Data Scientist
Vega IT | Novi Sad  ·  2023 – present

• Research and development with applications of Large Language Models in InsurTech
• Developing an agentic pipeline for knowledge extraction from documents
• Prompt engineering
• Python, PostgreSQL, HuggingFace, LangChain, Azure, Git

Researcher
Institute for Systems and Robotics | Lisbon  ·  2019 – 2023

• Research on task offloading in networks of machines
• Developing deep reinforcement learning agents in PyTorch
• Research in distributed optimization and ML models with applications in the animation industry

Researcher
Faculty of Sciences | Novi Sad  ·  2018 – 2019

• Developing models for distributed implementation of ML algorithms

Intern
BioSense Institute | Novi Sad  ·  2018

• Developing a classifier to recognize cultures planted in specific fields using satellite imagery

Education

PhD in Statistics and Stochastic Processes
Instituto Superior Tecnico | Lisbon  ·  2019 – present

• The curriculum covers machine learning, optimization, and statistics with a high demand for both theoretical and practical skills
• Thesis: ``Distributed optimization of bio-kinetic models based on large 4D sequences’’
• Solving large-scale optimization problems in facial animation, focusing on distributed optimization to reduce computational costs

MSc in Applied Mathematics – Data Science
Faculty of Sciences | Novi Sad  ·  2016 – 2018

BSc in Applied Mathematics – Mathematics of Finance
Faculty of Sciences | Novi Sad  ·  2013 – 2016

Skills

Technical Skills

• Python with PyTorch and TensorFlow
• Microsoft Azure Cloud
• MySQL and PostgreSQL
• PowerBI
• Git

Conceptual Skills

• Problem solving and critical thinking
• Machine learning and deep learning
• Data analysis
• Natural language processing and image processing
• Optimization
• Statistics

Publications

Clustering of the Blendshape Facial Model

A hybrid compartmental model with a case study of COVID-19 in Great Britain and Israel

High-fidelity Interpretable Inverse Rig: An Accurate and Sparse Solution Optimizing the Quartic Blendshape Model

Distributed Solution of the Inverse Rig Problem in Blendshape Facial Animation

Accurate and Interpretable Solution of the Inverse Rig for Realistic Blendshape Models with Quadratic Corrective Terms

A majorization–minimization-based method for nonconvex inverse rig problems in facial animation: algorithm derivation

Extreme Multilabel Classification for Specialist Doctor Recommendation with Implicit Feedback and Limited Patient Metadata

PeersimGym: An Environment for Solving the Task Offloading Problem with Reinforcement Learning

Refined Inverse Rigging: A Balanced Approach to High-fidelity Blendshape Animation

Talks

Big data processing for multivariate, noisy, and time-correlated data: human face reconstruction

July 01, 2019

Conference talk at ICIAM 2019, Valencia, Spain

MathYourAvatar

November 27, 2020

General public at MeetMeTonight, Italy (virtual)

Segmentation of the Blendshape Face Model

April 01, 2021

Conference Talk at 21st ECMI, Wuppertal, Germany (virtual)

3 Minute Thesis

May 01, 2021

Short presentation at 3MT2021,

Solving the Inverse Rig Problem for Highly-realistic Blendshape Models

July 02, 2021

Conference talk at EURO Conference, Athens, Greece

Clustering of the Blendshape Facial Model

August 01, 2021

Conference talk at EUSIPCO2021,

Solving the Inverse Rig Problem for Highly-realistic Blendshape Models

July 02, 2022

Conference talk at EURO Conference, Helsinki, Finland

High-fidelity Interpretable Inverse Rig

April 14, 2023

Conference talk at LYMC2023, Lisbon, Portugal

Distributed Solution of the Blendshape Rig Inversion Problem

April 14, 2023

Conference talk at International Convention Centre (ICC), Sydney, Australia