AT.
A crimson spiral galaxy among faint red nebulae

Sc. 01 — Opening

Italy / Open to new opportunities

Computer Engineer

AntonioTangaro

Computer Engineer and MSc student in AI & Robotics. I love algorithms, machine learning and the physics that makes them matter.

Explore work

Local time

Sc. 02 — About

Algorithms, learning machines and the physics in between.

I'm Antonio, a Computer Engineer finishing an MSc in AI & Robotics. I like the corners of computing where elegance and speed meet: an algorithm that drops a factor of n, a data structure that makes a hard problem obvious, a model that learns what nobody wrote down.

Download résumé
  1. 01

    Algorithms

    Graphs, dynamic programming and the quiet joy of a tight complexity bound.

  2. 02

    Machine learning

    From gradient descent to agents that learn by trial and error.

  3. 03

    Physics

    Forces, tyres and dynamics: models that have to obey the real world.

A white concrete tower with a single black slit, against a black sky
Still 02 — Behind the workPhoto: Carl Nenzen Loven, Unsplash License

Sc. 03 — Manifesto

Good engineering is invisible: the right abstraction, the right data structure, no more code than needed.

A white staircase seen through a narrow opening in a white wall
Still 03 — The quiet partPhoto: Todd Quackenbush, Unsplash License

I'm curious about how computing works all the way down: from voltages switching inside a circuit, to logic gates and automata, to the languages and programs we build on top, up to machines that learn from data instead of following rules.

I enjoy systems programming as much as mathematics, I care about tests and observability, and I would rather measure than guess. The best moment is when theory, code and hardware finally agree.

Skills: Python, PyTorch, Java, Rust, Elixir, C++, JavaScript, ROS 2, Docker, Kubernetes, Spring, PostgreSQL, Redis, Prometheus, Grafana, Linux, GitHub Actions, Typst

Sc. 04 — Work

Selected work

  1. 01 / Master's thesis

    Racing at the Limit

    A model-free reinforcement learning framework that learns to minimise lap time at the limit of handling on the Barcelona-Catalunya circuit. A nonlinear single-track model with Pacejka tyres, calibrated on a Porsche 718 Cayman GT4, runs as a vectorised Gymnasium environment for PPO training across 320 parallel environments. Policies complete 49 of 50 laps from standing starts, on average within 4.5% of the optimal lap computed with CasADi/IPOPT.

    Reinforcement Learning · PPO · Gymnasium · Vehicle Dynamics · Optimal Control · Python2026

    Photo: Bert van Dijk, CC BY-SA 2.0

    View code

  2. 02 / Research paper

    AoI & Crowdsensing

    A game-theoretic analysis of strategic participation in time-critical mobile crowdsensing, using Age of Information to measure freshness. Derives Nash and socially optimal participation, shows the gap scales as ln(N)/ρ, and designs Stackelberg subsidies that restore the social optimum.

    Game Theory · Age of Information · Crowdsensing · Optimisation2025

    Photo: Nicolai Berntsen, Unsplash License

    View code

  3. 03 / Game development

    Klotski

    The Klotski sliding-block puzzle in Java with LibGDX, running on desktop and Android, with a built-in solver and multiple difficulty levels.

    Java · LibGDX · Android2023

    Photo: Welt-der-Form, CC BY-SA 3.0

    Live demo View code

Also designedCinematic portfolio Minimal photographer portfolio

Sc. 05 — Experience

The story so far

  1. Jan 2024 — Oct 2024

    Software Engineering Intern — IoT & Embedded Systems

    SoundFood s.r.l.

    Designed and built an IoT distributed audio system for restaurants, giving every table its own independently controlled audio environment on Raspberry Pi Zero 2 W devices.

    • Architected the audio pipeline around Mopidy and Snapcast for synchronised multi-device playback with independent zones.
    • Developed the control layer and web interface in React and Redux, integrating Mopidy via WebSocket and Snapcast via JSON-RPC.
    • Optimised the embedded stack on DietPi, keeping CPU below 30% and memory around 150 MB in operation.

    Internship · Italy · Embedded Systems, IoT, Raspberry Pi, React, Redux, Python, WebSocket, JSON-RPC

Education

  • Sep 2024 — Dec 2026M.Eng in Computer Engineering — AI & RoboticsUniversity of Padova, ItalyThesis: Reinforcement Learning for Autonomous Racing at the Limit of Handling.
  • Sep 2021 — Sep 2024B.Eng in Computer EngineeringUniversity of Padova, ItalyThesis: Architecture of an audio streaming system on Raspberry Pi for controlling sonic microenvironments.

Languages

  • ItalianNative
  • EnglishFluent

Sc. 06 — Process

Every tenth matters.

From the first equation to the last measurement, the same care goes into modelling, code and results.

  1. 01

    Model

    Turn the problem into precise dynamics, constraints and objectives.

  2. 02

    Simulate

    Build fast, vectorised environments that scale to millions of decisions.

  3. 03

    Learn

    Train policies and systems that adapt instead of following a script.

  4. 04

    Measure

    Benchmark against optimal baselines until the gap is honest and small.