Skip to content
View AndreaTorti-01's full-sized avatar

Highlights

  • Pro

Block or report AndreaTorti-01

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
AndreaTorti-01/README.md

Andrea Torti

I obtained my Master's Degree in Computer Science and Engineering @ Politecnico di Milano in 2025 and I specialized in low-level programming and optimization, with some computer graphics, computer vision, robotics, embedded development and AI sprinkled here and there. I even wrote a lot of raw CUDA code and obtained a couple certifications from Nvidia!

During my academic career, I loved hands-on projects the most, where I could have control over every piece of the software I was developing and take pride in the results. Most of them were group projects, and that's where I discovered that I perform the best working with a team on tough problems and under strict deadlines. Some were developed on Windows, most on Linux, but I also daily macOS.

Here's a preview of some of my work, but there's always something more that's work in progress... always building, always learning!

CFD Software written from scratch entirely in C++! Can run on CPU (Even in parallel with OpenMP) or on your NVIDIA GPU!

On GPU, performance was the top priority, so I chose the structure of arrays approach. On CPU, arrays of structures is used for better code readability.

Yes, the performance was profiled with callgrind. Down to the single function call.

Did you think I wouldn't profile the GPU too?


Onto the next one. My master thesis, actually. Have you ever eaten instant noodles? Well, sometimes their packaging is not perfectly sealed...

Having only about a hundred or so profilometer scans, I had to make the computer do its magic to split the good ones from the bad ones. The obvious solution is machine learning, but it's not easy at all with a low sample count like that. I chose to pre-train my model on synthetic data generated with blender on packages like the one below:

After proper automatic pre-processing of each package profile...

The accuracy went to 89% with very good spot detection of defects! Mission accomplished!


Another one! The data you see is from an Arduino's motion sensors.

My friend chatgpt wrote some python software to cut 1-second samples...

And I trained a tensorflow deep learning model to recognize what the person wearing the arduino was doing! How cool is that? Oh and it runs ON the actual arduino!


Do you like trading? Here's a little trading game for you! Done over a weekend, needs some polish. At least it works (sometimes)


Oh yeah I took a Vulkan course at university. Here's a little game running on an engine... written from scratch.


Here you go: two static websites I made. First one: modern framework, about modern C++. Right in your face.

Github Pages Link

Second one with some personal recipes. VanillaJS technology. Yes I even like cooking! And yes you can read it too since it's bilingual. Now that I think about it I should do a travel blog too...

Netlify Link


Here goes the first thing I ever wrote in CUDA: Took an existing software that compresses molecule data (Smiles representation) and made it run on the GPU... 44 times faster! Oh and that's only because the disk couldn't keep up!

Pinned Loading

  1. Lattice-Boltzmann-Method Lattice-Boltzmann-Method Public

    Implementation of the Lattice Boltzmann Method in C++ and CUDA

    C++ 13 5

  2. MolCompPresentation MolCompPresentation Public

    This project implements a high-performance lossless codec for SMILES strings, designed for efficient compression and decompression. It's compatible with CUDA and AMD's HIP framework

  3. computer-graphics-project-2024 computer-graphics-project-2024 Public

    Forked from Comodaino/computer-graphics-project-2024

    C++

  4. HAEEAI_Falzoni_Torti HAEEAI_Falzoni_Torti Public

    Real-time football activity classification using embedded AI on Arduino Nano 33 BLE Sense with TensorFlow Lite

    C

  5. 2d-lbm-cuda 2d-lbm-cuda Public

    Cuda 1 1

  6. Trading-Simulator Trading-Simulator Public

    python minigame to simulate intraday trading

    Python 1