top of page
IMG_2858.HEIC

Gianmarco Scarano

M.Sc. in Artificial Intelligence & Robotics | "La Sapienza" University of Rome

Hey there! 
I am Gianmarco Scarano and I hold a Master's Degree in Artificial Intelligence & Robotics at "La Sapienza" University of Rome, in Italy!
I also earned my Bachelor's Degree in Computer Science and Technologies for Software Production from the University of Bari "Aldo Moro".

In my free time I like to produce music (mainly lo-fi and ambient), shoot cool pictures, play games, football and just code!

photo_5852990232385933358_y.jpg
photo_5852990232385933352_y.jpg
photo_5852990232385933354_y.jpg

Core Competencies

High experience in Python and Java programming languages & basic knowledge of C, C#, C++ and SQL/NoSQL.

Able to build and optimize existing ML/DL models through SOTA algorithms.

Proficient knowledge of Tensorflow, Keras, PyTorch & Sklearn libraries along with high knowledge of the Git library.

My EDUCATIONAL CAREER

Sept 2013 - June 2018

Senza titolo_edited.png

I.T.E.S. F.M. GENCO

High School Diploma, Economia aziendale/manageriale

Sept 2018 - Apr 2022

channels4_profile_edited.png

Università degli studi di Bari

Laurea triennale in Informatica e Tecnologie per la Produzione del Software (ITPS)

- Vision Transformer (ViT) for Crowd Counting purposes
- Neural Network for predicting the speaker's gender
- Neural Network for predicting the speaker's emotion
- Developed an Arcade Game called "Arkanoid" for Android (through Android Studio)
- Java programming language
- Python programming language
- C programming language
- SQL
- OO Paradigms
- Cyber Security activities with tests on VulnHub
- MVC Patterns
- Usability Tests & UI Examples 
- Code Quality Inspections (SonarCloud, Maven, etc.)
- Spring Framework (Redmine, GANTT, etc.)
- GitHub first approaches & insights
- B2 First Certificate

logo_RGB_negative.png

Sept 2022 - Jan 2025

Sapienza Università di Roma

Master Degree, Artificial Intelligence & Robotics

- Decision Trees, Bayesian Learning, SVM with various kernels (Poly, RBF, etc.)
- Linear Classification & Regression tasks
- Neuroengineering basics (Neural Encoding/Decoding, Brain Networks, EEG, BCI)
- Deep Neural Networks (NN fundamentals, CNN, Gradient computation, Transformers, GANs, VAEs, CVAEs) using PyTorch & TensorFlow/Keras
- Computer Vision tasks (OpenCV, Epipolar Geometry, Camera Calibration, Segmentation and Detection, Multimedia Forensics)
- Robot Programming (C++, ROS support)
- Reinforcement Learning (SARSA-λ, Q-Learning, DDQN, Actor-Critic, A2C, PPO)
- Natural Language Processing (Similarity, Embeddings and Multimodality)
- Information Retrieval through DSI (using LLMs such as T5, Flan-T5, SwitchTransformers, Mixture-of-Experts, LoRA and QLoRA)
- SQL/NoSQL databases (MongoDB with Compass & NoSQLBooster)
- Robotics (Direct/Inverse/Differential Kinematics, Statics, Force, Dynamic Models & Trajectories)
- Autonomous Systems (Mobility, Trajectories & Motion Planning)

Competenze: Artificial Intelligence · Machine learning · Deep Learning · Computer Vision · Natural Language Processing · Robotics

MTMCRetrieval.png

Vehicle Re-Identification using ResNet
and Spatio-Temporal constraints

Designed and implemented a reidentification (Re-ID) system which accurately tracks a target vehicle across complex camera networks using advanced CNN Feature Extraction protocols and sophisticated Deep Learning algorithms for Multi-Camera Trajectory Linking.

Senza titolo-1.png

Neural Inverted Index for
Information Retrieval

The project aims to develop a unified model that replicates an index-based document retrieval system, returning a ranked list of document IDs for a given query, introducing a novel approach to DSI models.

Screenshot 2025-02-08 110057.png

SoftKick

Implement Proximal Policy Optimization (PPO) algorithm from Reinforcement Learning literature in the Psyonix game 'Rocket League' for an efficient kick-off strategy.

Project showcase

And some final images.. 

my experiences

smart_i_logo.jpeg

Designed and implemented a reidentification (Re-ID) system which accurately tracks a target vehicle across complex camera networks using advanced CNN Feature Extraction protocols and sophisticated Deep Learning algorithms for Multi-Camera Trajectory Linking.

1631300751801.jpeg

4 weeks course (120h in total) aimed at learning the English Tourism and Marketing background, through an English language level equivalent to Cambridge B2 (First).

1631345726746.jpeg

4 weeks course, entirely in English, for PET (B1 Preliminary) preparation. Very useful experience which gave me the first opportunity to connect with people in an international context.

AI Engineer Intern

Smart-I Srl · Stage

giu 2024 - dic 2024 · 7 mesi

Roma, Lazio, Italia

Upper-Intermediate English course for Tourism, Business & Marketing
Atlas Language SchoolAtlas Language School
ago 2018 - ago 2018 · 1 mese
Dublino, Irlanda

PET Preparation
Cambridge Melchior College (CMC)
set 2017 - set 2017 · 1 mese
Cambridge

Licences and certifications

B2 First
Università degli Studi di Bari
Data di rilascio: ago 2021

channels4_profile_edited.png

PET Preliminary
Cambridge Melchior College (CMC)
Data di rilascio: nov 2017

1631345726746.jpeg

Contact me!

bottom of page