I'm studying Artificial Intelligence & Data Analytics at the University of Trieste. I'm interested in computational complexity, formal methods, and systems programming. I like implementing things to understand how they work, then finding ways to check that they do what I expect.
-
Tiling Foundry — implements the Yang–Zhang reduction from CM1-in-3 SAT to a fixed 23-tile Wang system, with reference and optimized C solvers, independent Boolean and Wang Z3 checks, and witness verification. Reproducible run dossiers keep results, traces, figures, and PDF reports together. Read the walkthrough.
-
Wang Studio — a browser editor for tiling experiments on square, triangular, and hexagonal grids. I use it to construct regions and tilesets, set boundary constraints, and explore solutions with Tiling Foundry's C solvers compiled to WebAssembly. Try the editor.
-
TensorForth — a C17 interpreter for a Forth-like tensor language, with OpenMP, reference-counted tensor storage, tests, and Valgrind checks.
-
ADS — a growing C library where I implement data structures and graph algorithms from scratch, focusing on understanding their implementation and trade-offs.
-
Cantina DB — a relational database and management app for winery inventory, with trigger-maintained stock, stored procedures, role-based access, and a Streamlit demo. Try the demo.
-
FlowScore — explores credit scoring for gig workers. I designed the architecture, synthetic-data engine, and shock scenarios to compare static credit scores with debt outcomes over time.
-
KaleidoSpeech — a team proof of concept built for a hackathon in Hangzhou, exploring pronunciation training through browser audio, phoneme alignment, and visual feedback.
PAP Render is a pixel-art renderer in Python and NumPy.
I've also worked on team hackathon projects: insurance coverage recommendations with Generali, demand forecasting and supply-chain optimization with Zucchetti, and e-bike energy optimization at EESTech.
Gelsomino Vine is an early experiment toward a local-agent system. I'm working on the protocol layer: deterministic messages, causal ordering, and conformance tests.