Hi there, I'm Piermichele!
Curious, collaborative and driven by continuous learning, I've always been fascinated by understanding how technology works and how it can be used to create meaningful impact. This passion led me to pursue a career in Computer Science, where I discovered a particular interest in AI, software development, and the intersection between technology and innovation. I enjoy exploring both theoretical and practical aspects of computing, from designing intelligent systems to developing software solutions that address real-world challenges.
I am particularly interested in research and innovation that combine AI, data, and knowledge representation to solve complex problems and create positive societal impact.
In my free time, I enjoy traveling, discovering new cultures, attending technology-related events, and staying up to date with the latest developments in AI and technology in general.
I build AI systems end to end: retrieval pipelines, agentic workflows, LLM integration. I don't stop at a live demo — I take them to production. It's my day job at Axians and the subject of my PhD.
RDF, ontologies and knowledge graphs are daily working tools for me. A real-world example: at Axians I used them together with AI systems to improve the processes of several business units by 85% across corporate projects.
I model a domain's real concepts and rules so machines can reason over them: that's the power of conceptual modeling and domain-specific modeling languages. It is also the core of my PhD at UPV.
I follow efficient methodologies across the whole software lifecycle, from gathering requirements through to release.
From idea to something you can click, quickly. Short feedback loops beat long specifications, so decisions get made on a working artifact instead of on paper.
I work at the current edge and stay there: agentic frameworks like LangGraph and LlamaIndex, Knowledge Graphs with LLMs, Claude Code in the daily loop. Publishing research keeps me honest about what is genuinely new.
Papilla: the menu that remembers you. A full-stack platform for restaurants where guests scan a QR code at the table, browse the digital menu, order and pay from their phone, and get suggestions based on their tastes. Staff run orders, kitchen and bar displays, tables, payments and analytics from a single panel. Built with Next.js, React, Supabase (PostgreSQL) and Stripe Connect.
Next.js
SupabaseThis website is a responsive, accessible portfolio with a terminal aesthetic, built with Next.js and Tailwind and deployed on Vercel.
Next.js
TypeScript
Tailwind CSS
VercelMaster's thesis project focused on improving the formation of research consortia for collaborative projects. The solution introduces an Agentic Graph RAG approach that combines Knowledge Graphs and Large Language Models to identify and recommend suitable research collaborators based on expertise, research interests, and project requirements. The system provides explainable and context-aware recommendations, demonstrating the potential of Neuro-Symbolic AI techniques to support research collaboration and consortium formation.
Python
GraphDB
Docker
Streamlit
Langchain
LlamaIndexTARNAS is a software solution designed to simplify the translation of RNA secondary structure files between different formats. It eliminates the need for manual conversions, reducing both time and errors when working with heterogeneous RNA datasets. TARNAS also provides data-cleaning features to remove unnecessary metadata from files. The tool is available as both a standalone application (GUI and CLI) and a web application, making RNA structure analysis more accessible and efficient for researchers.
Java
JavaFX
Maven
ANTLR
Picocli
Springboot
AngulariModuleBuddy is a hybrid AI-powered study planning assistant that combines Knowledge Graphs, semantic search, and Large Language Models to generate personalized academic plans for postgraduate students. By leveraging the ESCO ontology and Retrieval-Augmented Generation (RAG), it recommends courses aligned with students' experience and career goals, organizes them into multi-semester study plans, and provides tailored recommendations with explainable insights.
PythonStreamlitSupabase
Neo4jLangchainLlamaIndexA simple application intended to provide a calculator of a person's Italian tax code. The objective of this calculator is to take as input a person's basic data such as surname, first name, date and place of birth and then output the Italian tax code.
PythonGEDCOM is a genealogy data processing tool designed to parse and analyze genealogical information stored in the GEDCOM format. Built using ANTLR4 for lexer and parser generation and Java for semantic processing, the system enables efficient parsing of GEDCOM files or streams and supports the computation of family relationships, including ancestor and descendant trees, based on user requests. It provides a structured and automated approach to genealogical data analysis and exploration.
JavaANTLRCasotto is a full-stack web application for managing a smart beach resort, developed with Spring Boot and Spring Security on the backend and Angular on the frontend. The platform enables customers to book beach umbrellas, order food and drinks directly from their spot, and register for recreational activities. It also provides management tools for staff and administrators, including order handling, sales tracking, personnel management, and service catalog administration. All data is securely stored in a database to ensure reliability and persistence.
Java
MySQLSpringbootAngularLOGO is a Java-based project developed for an Advanced Programming course. The application parses and executes LOGO commands from text files to generate geometric figures on a graphical canvas. Built with Gradle, it demonstrates concepts such as language interpretation, command execution, and graphical rendering, allowing users to visualize shapes created through LOGO instructions.
JavaJavaFX
Gradle
Zürich, Switzerland
Python
Azure Functions
RDF
Milan, Italy
Zürich, Switzerland
PythonAzure FunctionsRDF
Olten, Switzerland
Olten, Switzerland
Java
Microsoft Power Apps
Microsoft Power Automate
Microsoft Sharepoint
Camerino, Italy
LaTeXOlten, Switzerland
PythonLangchainLlamaIndexNeo4j
San Benedetto del Tronto, Italy
Java
Acquaviva Picena, Italy
C#
Italian (Native)
English (Professional working proficiency)
German (Learning)PythonCJavaJavaFX
RTypescriptAzure FunctionsSpringboot
HTML
CSSAngularStreamlitLlamaIndexLangchainLangGraph
Claude CodeMySQL
Azure CosmosDBGraphDBNeo4jMicrosoft Azure (Resource Groups, Function apps, App Services)
Microsoft AI FoundryRDF
OllamaDocker
Git
GitHub
GitLab
BitbucketVercel
UPV - Polytechnic University of València
València, Spain
thesis: It'll be added sooner or later :)
FHNW - University of Applied Sciences and Arts Northwestern Switzerland
Olten, Switzerland
thesis: "A Hybrid AI Approach for Recommending Collaborators in Research Projects". This thesis presents an Agentic Graph Retrieval-Augmented Generation approach to improve the recommendation of research collaborators. By integrating Knowledge Graphs and Large Language Models, the system delivers contextual, explainable and personalized suggestions aligned with researchers' expertise and project needs. The method was developed using Design Science Research and evaluated with GPT-4o, showing promising results in addressing limitations of traditional collaboration tools.
Unicam - Univeristy of Camerino
Camerino, Italy
thesis: "A Hybrid AI Approach for Recommending Collaborators in Research Projects". This thesis presents an Agentic Graph Retrieval-Augmented Generation approach to improve the recommendation of research collaborators. By integrating Knowledge Graphs and Large Language Models, the system delivers contextual, explainable and personalized suggestions aligned with researchers' expertise and project needs. The method was developed using Design Science Research and evaluated with GPT-4o, showing promising results in addressing limitations of traditional collaboration tools.
Unicam - Univeristy of Camerino
Camerino, Italy
thesis: "TARNAS: TrAnslator for RNA Secondary structure formats". As part of a group thesis, I co-developed TARNAS, a software tool for translating RNA secondary structure formats. The project applied languages and compilers theory to bioinformatics, addressing inconsistencies across existing file representations. TARNAS supports six common formats and features automatic format recognition (parser-based), customizable translation and header-cleaning options.
F. Di Petta, P. Rosati, P. Hierro Canchari, M. Quadrini, L. Tesei
Computational Intelligence Methods for Bioinformatics and Biostatistics (bioRxiv preprint) 2026
Docker
P. Rosati, E. Laurenzi, M. Quadrini
International Conference on Society 5.0 2025
M. Quadrini, P. Hierro Canchari, P. Rosati, L. Tesei
International Journal of Molecular Sciences 2025
M. Spahic-Bogdanovic, H. Witschel, D. Porumboiu, P. Rosati, P. Hierro Canchari, M. Kostic
HybridAIMS 2025 @ CAiSE 2025
M. Quadrini, P. Hierro Canchari, P. Rosati, L. Tesei
Computational Intelligence Methods for Bioinformatics and Biostatistics 2024
# Got a project, a collaboration, or just want to say hi? Ping me. Whether we meet for a coffee in Switzerland or sync up online, my inbox is always open.
# Looking for my CV? It's available on request, just send me an email and I'll get it to you.