MSc student in AI & Human-Centered Computing.Ericsson intern, benchmarking Post-Quantum TLS 1.3 across 105,000 handshakes.
MSc student in AI & Human-Centered Computing (University of Genoa) with coursework in Machine Learning, Deep Learning, Operations Research and Automated Reasoning. Ericsson intern: designed and statistically analyzed a 105,000-handshake benchmark of Post-Quantum TLS 1.3. Built a CNN defect detector (PyTorch, ResNet18) robust to real-world perturbations and an LLM-powered Android app as Bachelor's thesis.
Graduate program combining Machine Learning and Deep Learning with Operations Research, Automated Reasoning and Human-Computer Interaction.
Coursework: Algorithms & Data Structures · Computer Networks · Control Systems · Embedded Systems · Databases. Thesis: CarSpot - LLM-assisted Android car marketplace.
Measured what Post-Quantum key exchange really costs in TLS 1.3 handshakes, across realistic network conditions.
Classical ML vs Deep Learning on 7,300+ industrial casting images: SVM, Decision Tree and Random Forest (GridSearchCV, 3-fold CV) against ResNet18 transfer learning in PyTorch (frozen vs full fine-tuning, augmentation, early stopping); train/test leakage removed via MD5 deduplication. ResNet18 reached 99.8% balanced accuracy; under a stress test simulating production conditions (rotation, shift, brightness) classical models collapsed to ~50% while ResNet18 held 91%+.
LLM-assisted Android car marketplace. Integrated an LLM assistant (OpenAI API) with domain-specific prompt engineering, handling 500+ conversational queries in testing. Real-time Firebase backend (Auth, Firestore, Storage) with <300 ms median query latency via indexed collections and pagination.
Multilingual tourism platform: 15+ pages in 5 languages, 300+ audio-guide assets, 500–1,000 monthly visitors. Built and deployed full stack.
Turn-based 2D strategy game in C++ (Unreal Engine 5) with two bot behaviours driven by minimax decision logic. Interactive UI with 2 unit types. Graded full marks.
Company website for Nova42, Rome: 5–7 production pages shipped over 12 months. Responsive front-end (HTML/CSS/JavaScript) and PHP back-end.
Web platform built with and adopted by IIS Calvino staff, digitizing PDP workflows for 50+ students and reducing paper-based bureaucracy for families and educators.