Projects, Achievements & Ideas, Some Cool, Most Lame
GDSC AI Hack
First place at the annual AI Hackathon in the Finance Track. Our team developed a PoC for a fantasy game for real world stocks to simplify learning and decrease financial illiteracy.
Learn More
Bachelor Thesis
Titled "Beyond Hallucination: Enhancing AI Content Creation with Verifiable Sources Using Retrieval-Augmented Generation" my thesis was supervised by Prof. Emanuele Borgonovo and explores RAG and trust in AI-generated content.
View Code on GitHub
Neural Network
Implemented from scratch a feedforward neural network in Python using only NumPy. The project included forward and backward propagation, training with gradient descent, and evaluation on standard datasets.
View Code on GitHub
Atari Breakout with Reinforcement Learning
Implemented an Atari Breakout game (both front and backend) with a human player and an AI agent mode. The latter was trained using reinforcement learning with monte carlo and exploring starts.
View Code on GitHub
Fantasy Soccer DApp
Developed a decentralized application (DApp) for fantasy soccer using Solidity and JavaScript. The project focused on security, role-based-authentication, token generation and draft mechanism design.
View Code on GitLab
Skipass Demand Prediction using KGs
Full Stack web application powered by a knowledge graph backend in Neo4J using data from the open weather portals of the North Italian regions. The goal of the project was to apply Knowledge Graph Embeddings such as TransE and RotatE for graph completion and link prediction and prove its real life applicability, e.g. in the winter entertainment industry.
View Code on GitHub
Basic Chooser iOS app: Tocca
A basic iOS app mimicking Chooser, to track hand gestures by multiple people and randomly select winners and losers or form teams.
View Code on GitHub