Federico Gianno

Open reference implementation for a single-agent system with tool routing, history compaction, and shared state across multiple frontends.

An AI player for the 2048 game, using the expectimax algorithm combined with heuristic evaluation to select the best future moves. The algorithm explores possible moves, averages over random tile spawns, and scores each resulting board to pick the best action.

Educational resource to understand how ransomware works under the hood by demonstrating the cryptographic principles and workflows commonly employed by this malware.

LLMs prompt augmentation with Retrieval Augmented Generation (RAG) by integrating external custom data from a variety of sources (e.g. text files, web pages, PDFs, etc.) allowing chat with such documents.

A thread safe in-memory key-value cache suitable for single instance microservices. Any object can be stored with a given duration or no expiration time at all. Since the cache is thread safe, it can be safely used by multiple goroutines.

League of Legends fight tool made in GoLang that allows you to download champions data, make them fight, and get the best combination of spells based on the time (in seconds) it takes to kill the enemy champion.

Android food delivery (Customer, Restaurant and Rider) APP:
- Customer app: search restaurants and manage the orders.
- Restaurant app: allows the restaurant owner to advertise the daily menu offer (e.g. create and maintain the restaurant profile, publish\edit the daily offer, show/manage the list of current reservations, etc.).
- Rider app (deliveryman): allows the riders to see the orders and access both the restaurant and the delivery address.

Food management home assistant app for Google Assistant or Cloud Speech-to-Text service. It uses natural language processor and it is able to connect to the Google Assistant or Cloud Speech-to-Text service.

Hungry Student APP has been developed in 24h during the "Like@Home" hackathon organized by Mu Nu Chapter of IEEE in Turin on 30-31th March 2019.

Thanks to this assistant we got the first place.

Convolutional Neural Network (CNN) implementation from scratch in PyTorch on CIFAR 100 image dataset.

The code implements a basic NN and CNN, the data loading, the training phase and the evaluation (testing) phase.

State-of-the-art research in order to study and design data models for Network Function Virtualization (NFV) and Software Defined Networking (SDN) architectures - associated with PoliTO Netgroup (Computer Networks Group at Politecnico di Torino).

- State-of-the-art research in the NFV and SDN network architectures.
- Research and development of a standard data-format (described by means of an XML schema) for the representation of all the most relevant information in the NFV and SDN… Show more

State-of-the-art research in order to study and design data models for Network Function Virtualization (NFV) and Software Defined Networking (SDN) architectures - associated with PoliTO Netgroup (Computer Networks Group at Politecnico di Torino).

- State-of-the-art research in the NFV and SDN network architectures.
- Research and development of a standard data-format (described by means of an XML schema) for the representation of all the most relevant information in the NFV and SDN context.
- Implementation of Java RESTful web service (Maven, JUnit, TomCat and GlassFish Server) that allows to store and retrieve the NFV/SDN information. Show less

1. PCA applied on images and Naive Bayes Classifier to classify them.

It is showed what happens if different principal components (PC) are chosen as basis for images representation and classification. Then, the Naive Bayes Classifier has been choosen and applied in order to classify the image.

2. Validation, cross validation and grid search with multi class SVM.

Given a set of training examples, each marked as belonging to one or the other of two categories, an SVM training… Show more

1. PCA applied on images and Naive Bayes Classifier to classify them.

It is showed what happens if different principal components (PC) are chosen as basis for images representation and classification. Then, the Naive Bayes Classifier has been choosen and applied in order to classify the image.

2. Validation, cross validation and grid search with multi class SVM.

Given a set of training examples, each marked as belonging to one or the other of two categories, an SVM training algorithm builds a model that assigns new examples to one category or the other, making it a non-probabilistic binary linear classifier. Show less

A system made up of ESP32 devices and a server that acts as a collector of the information collected and allows its processing and display.

The firmware of the ESP is written with the purpose of sniffing probe request packets sent by smartphones that are looking for a Wi-Fi connection. From each sniffed packet, some information is extracted, and after each minute, they are sent to a server that processes and stores them in a SQL database.

It is possible to see the processed… Show more

A system made up of ESP32 devices and a server that acts as a collector of the information collected and allows its processing and display.

The firmware of the ESP is written with the purpose of sniffing probe request packets sent by smartphones that are looking for a Wi-Fi connection. From each sniffed packet, some information is extracted, and after each minute, they are sent to a server that processes and stores them in a SQL database.

It is possible to see the processed information, like the real-time location of the smartphones, smartphone frequency, etc., through a GUI. Show less

The purpose is to assign time-slots and resources to the meetings so as to satisfy the constraints as much as possible. The problem has been solved with metaeuristic algorithms (i.e. Graph Coloring, Tabù Search and Simulated Annealing).

Python telegram BOT designed to facilitate the study of students.

WeStudy has been developed in 24h during the "Hello B0t!" hackathon organized by Mu Nu Chapter of IEEE in Turin the 11th-12th November 2017.

Thanks to this bot, my team and I got first place in the competition.

Generate customised wordlists for penetration testing practice (e.g., brute force attack, dictionary attack, etc.)

Report about Twitch application (a streaming service, twitch.tv) analysing the various protocols, how connection and data are handled/managed (e.g. login phase, chat, streaming, etc.) and decrypting data.

Prepare your wedding at best by seating your guests in the best possible way based on each guest's interaction with the others, by calculating the average mood of the room and assigning each guest a specific seat.