Software Developer @ Adeptic Reply · MSc @ UNICAM & Reykjavik University
Software Developer @ Adeptic Reply · MSc @ UNICAM e Reykjavik University
I'm a software developer at Adeptic Reply in Turin, working on IPCEI-CIS — a European initiative shaping the next generation of multi-provider Cloud/Edge infrastructure.
I hold a double MSc in Computer Science from the University of Camerino and Reykjavik University, with a specialisation in Machine Learning and NLP. My thesis applied deep reinforcement learning (CLSTM-PPO) to algorithmic financial trading.
Outside the terminal, you'll find me climbing — sport climbing is my way of keeping the mind clear.
Sono un software developer ad Adeptic Reply a Torino, dove lavoro su IPCEI-CIS — un'iniziativa europea che plasma la prossima generazione di infrastrutture Cloud/Edge multi-provider.
Ho conseguito il doppio MSc in Computer Science tra l'Università di Camerino e la Reykjavik University, con specializzazione in Machine Learning e NLP. La mia tesi applica il deep reinforcement learning (CLSTM-PPO) al trading finanziario algoritmico.
Al di fuori del lavoro, puoi trovarmi a praticare arrampicata sportiva — è il mio modo di mantenere la mente libera.
Working on the IPCEI-CIS European project, developing cloud-native services and APIs for a federated multi-provider Cloud/Edge continuum. Key contributions:
Lavoro sul progetto europeo IPCEI-CIS, sviluppando servizi cloud-native e API per un continuum Cloud/Edge multi-provider federato. Contributi principali:
Developed native iOS applications, working on Barterchain's home page from scratch. Gained hands-on experience with XCode and Swift.
Sviluppo di applicazioni iOS native, con lavoro sulla home page di Barterchain da zero. Esperienza pratica con XCode e Swift.
Implemented a Business Management Software using Blazor and Visual Studio 2019, applying the MVVM design pattern to build interactive dashboards.
Sviluppo di un gestionale aziendale con Blazor e Visual Studio 2019, applicando il pattern MVVM per la realizzazione di dashboard interattive.
Focus on Machine Learning and Natural Language Processing. Recipient of the highly competitive Double Degree Scholarship (10 positions awarded for academic excellence).
Specializzazione in Machine Learning e Natural Language Processing. Borsa di studio Double Degree (10 posti, merito accademico).
Software Development and Technologies curriculum. Thesis: "Time Series Forecasting in Automated Financial Trading" — CLSTM-PPO deep reinforcement learning applied to algorithmic trading.
Curriculum Software Development and Technologies. Tesi: "Time Series Forecasting in Automated Financial Trading" — deep reinforcement learning CLSTM-PPO applicato al trading algoritmico.
Thesis: "Study and analysis of the Single Shot Multibox Detector algorithm applied to the cataloguing of a dataset of historical images".
Tesi: "Studio e analisi dell'algoritmo Single Shot Multibox Detector applicato alla catalogazione di un dataset di immagini storiche".
MSc thesis. CLSTM-PPO deep reinforcement learning architecture for predicting financial market trends, combining statistical techniques and deep learning.
Tesi MSc. Architettura CLSTM-PPO di deep reinforcement learning per la previsione dei mercati finanziari, combinando tecniche statistiche e deep learning.
Seq2Seq model for translating aphasic sentences (Broca's aphasia) into grammatically correct English. NLP course at Reykjavik University.
Modello Seq2Seq per tradurre frasi afasiche (afasia di Broca) in inglese corretto. Corso NLP alla Reykjavik University.
Dynamic malware analysis and classification using Cuckoo Sandbox. Final project for the MLCS course at Reykjavik University.
Analisi e classificazione dinamica di malware con Cuckoo Sandbox. Progetto finale MLCS alla Reykjavik University.
Decision Tree (99.87% accuracy) and Random Forest (81.79%) classifiers with thorough preprocessing for cybersecurity threat detection. Built with scikit-learn and pandas.
Classificatori Decision Tree (99.87%) e Random Forest (81.79%) con preprocessing avanzato per la rilevazione di minacce informatiche. Sviluppato con scikit-learn e pandas.
Stacking ensemble ML model achieving 0.999 accuracy — 0.001 above state of the art. Group project for the Research Methodologies course at Reykjavik University.
Modello ensemble ML con stacking: 0.999 di accuracy, superiore allo stato dell'arte. Progetto di gruppo per il corso Research Methodologies alla Reykjavik University.
Evaluation of Naive Bayes, KNN, and Logistic Regression spam classifiers. Naive Bayes achieved 95.77% accuracy, 99.17% precision, AUC 99.79%.
Valutazione comparativa di Naive Bayes, KNN e Logistic Regression. Naive Bayes: 95.77% accuracy, 99.17% precision, AUC 99.79%.
REST API for querying JanusGraph with a Cassandra backend. Built for the Technologies for Big Data Management course at UNICAM.
REST API per interrogare JanusGraph con backend Cassandra. Realizzato per il corso Technologies for Big Data Management a UNICAM.
Face and stamp recognition in historical images using SSD (Single Shot MultiBox Detector). Trained on local museum data.
Riconoscimento di volti e francobolli in immagini storiche tramite SSD. Addestrato su dati di un museo locale.
Java library implementing graph algorithms including Dijkstra and Bellman-Ford. Course project at UNICAM.
Libreria Java con algoritmi su grafi (Dijkstra, Bellman-Ford). Progetto del corso ad UNICAM.
Java framework for card games, including an implementation of the Italian game Briscola. Developed at UNICAM.
Framework Java per giochi di carte con implementazione della Briscola. Sviluppato a UNICAM.
Full work history, education, and skills in PDF format.
Storico completo, formazione e competenze in formato PDF.