Rome, Italy — Software Engineer / AI Engineer

Andrea
Barontini

Game Development Machine Learning Theoretical Physics PhD Leonardo S.P.A.

I build things at the intersection of deep theory and applied craft — from neural-network PDF fits at the LHC to indie games shipped on itch.io.

View Game Projects View AI Projects Publications
Scroll
01 —

Game Dev

Screenshot
Poor Astronaut
A space survival game built in Unity. Navigate a series of increasingly challenging levels to escape the shuttle!.
UnityC#2DPlatform
↗
Screenshot
Space Shuttle
2D side-scrolling arcade game. Dodge debris across escalating difficulty waves.
UnityC#2DArcade
↗
Screenshot
Royal Run
A 3D procedurally generated runner. Dodge obstacles as you escape the castle.
UnityC#3DRunner
↗
Screenshot
Space Escape
A 3D arcade-style shooter. Destroy enemy turrets using robot laser fire.
UnityC#3DRunner
↗
02 —

AI / ML Projects

Tool · Open Source
Pineline
An industrialisation framework for high-energy theory predictions. Harmonises inputs, standardises grid outputs, and tracks metadata — enabling large-scale PDF fit automation that previously required weeks of manual computation.
PythonNNPDFQCDPublished
2023
arXGH
Tool · Open Source
Pineline
An industrialisation framework for high-energy theory predictions. Harmonises inputs, standardises grid outputs, and tracks metadata — enabling large-scale PDF fit automation that previously required weeks of manual computation.
Python NNPDF QCD Published
2023
arX GH
Research · Neural Networks
MHOU in PDF fits
Missing Higher Order Uncertainties included in NNLO Parton Distribution Functions via a theory covariance matrix formalism. Demonstrates improved perturbative consistency and moderate reduction of PDF uncertainties.
TensorFlow NNPDF 4.0 Covariance Matrix Published
2024
arX
Research · Deep Learning
N³LO Parton Distributions
Extending NNPDF4.0 to approximate next-to-next-to-next-to-leading order via neural network fits. Includes N3LO splitting function approximations, DIS corrections, and FONLL general-mass scheme at O(αs³).
Neural Networks QFT LHC Published
2024
arX
Thesis Supervision · Explainability
Explainable AI in PDF Fits
Supervised three master theses on diagnostic tools for understanding internal neural network structure — closure tests, activation analysis, and feature attribution inside NNPDF architectures.
XAI Interpretability Thesis Supervision
2022–24
03 —

Publications

— - 04
The Path to N³LO Parton Distributions
arXiv:2402.18635 · 2024
Extending NNPDF4.0 to aN3LO with full uncertainty treatment via theory covariance matrix formalism. Impact on LHC Higgs and Drell-Yan cross-sections assessed.
arX
— 04
The Path to N³LO Parton Distributions
arXiv:2402.18635 · 2024
Extending NNPDF4.0 to aN3LO with full uncertainty treatment via theory covariance matrix formalism. Impact on LHC Higgs and Drell-Yan cross-sections assessed.
arX
— 03
MHOU on NNLO Parton Distributions with Percent Accuracy
arXiv:2401.10319 · 2024
Missing higher order uncertainties included in NNPDF4.0 through scale variation. Improved consistency and reduced PDF uncertainties at NNLO demonstrated.
arX
— 02
Photons in the Proton: Implications for the LHC
arXiv:2401.08749 · 2024
QED-evolved NNPDF4.0 set including photon PDF via LuxQED formalism. Electroweak corrections reach 5% in vector boson pair production at large invariant mass.
arX
— 01
Pineline: Industrialization of High-Energy Theory Predictions
arXiv:2302.12124 · 2023
Automation framework for large-scale theory prediction computation. Harmonised inputs, standardised grid outputs, and reproducible metadata across the NNPDF pipeline.
arX
04 —

About

I am a Software engineer with a Ph.D. in Theoretical Physics, where I conducted research on neural networks as part of the NNPDF Collaboration. My background combines scientific research, machine learning, and software engineering.

I currently work as a Software Engineer at Leonardo S.p.A., where I primarily focus on backend development in Python. I also lead a small development team on a full-stack project, contributing across both backend and frontend components while driving technical delivery and collaboration.

Outside of my professional work, I am passionate about game development. I have completed numerous courses covering topics from game programming fundamentals to advanced development techniques, and I actively develop personal projects using both Unity and Unreal Engine. Through these projects, I continue to expand my skills in software architecture, gameplay programming, and interactive experiences.

Physics & Math
QFT QCD PDF fitting Advanced modeling Advanced linear algebra
Code
Python C# C++ NumPy TensorFlow DesignPatterns LaTeX
Game Dev
Unity 3D / 2D Unreal Engine Game Design
Tools
Git Github Workflows JetBrains Rider Visual Studio
10+
Papers published
15+
Talks & seminars
3
Games shipped
3
Theses supervised
Currently
Python Software Engineer
, Leonardo SpA · Rome · 2024 – present
Contact
andreabarontini97@gmail.com ↗