AI researcher and engineer with 12+ years building intelligent systems — from algorithmic fairness and recommender systems to autonomous agents and enterprise LLM platforms. Based in Italy, operating at the frontier of industrial AI.
I'm a PhD computer scientist and entrepreneur, currently serving as CTO & Co-founder of Neural Factory — an AI platform company accelerating Italian manufacturing SMEs through intelligent automation and data-driven decision-making.
My doctoral research at the University of Cagliari focused on using Machine Learning to optimize e-coaching platforms — combining recommender systems, fairness algorithms, and dropout prediction. Years before "responsible AI" became a trend.
Before co-founding Neural Factory, I led AI initiatives at FairMind, designed data-driven marketing systems for Stellantis at Stackhouse, and built ML products that helped apps scale to millions of users across Italy, Spain, and Morocco.
Production-ready LLM systems — RAG pipelines, AI agents, orchestration layers, and guardrails that reduce hallucinations in complex enterprise contexts. Contributed open-source models on HuggingFace.
Published researcher in algorithmic fairness applied to recommendation. First in literature to introduce fairness-of-exposure algorithms for athletic e-coaching platforms.
Three-tier AI architecture for manufacturing SMEs: knowledge graphs + conversational analytics + autonomous agents. Real clients, real production deployments.
Serverless microservices on AWS, CI/CD pipelines, disaster recovery planning, and scalable ML SaaS products deployed across Europe.
Leading the technical vision of an AI platform company serving Italian manufacturing SMEs. Designed a three-tier AI system combining knowledge graphs, conversational analytics, and autonomous agents. Strategic partnerships with Alphabridge and Easynet Group.
Led R&D of a generative AI platform for software development acceleration. Designed production-ready LLM systems for code generation, built AI agents for complex problem-solving, developed model orchestration layers, and implemented guardrails for security and compliance. Contributed open-source models to HuggingFace including Minerva-3B-Instruct and multilingual adaptations. Provided technical mentoring to junior engineers.
Led design and deployment of data-driven marketing AI for Stellantis across Europe. Built an Audience Classification system, an Accessories Recommendation engine, and a Churn Prediction system — all deployed for multiple Stellantis brands simultaneously.
+20% conversion rate Multi-brand, multi-marketReporting to CTO. Led design of web applications, recommender systems, and LLM-based digital assistants. In collaboration with Arkys, developed a traceability and optimization solution for the waste management cycle. Deployed multiple Information Retrieval and Recommendation engines in production.
Led design and development of ML SaaS products for U4FIT, a digital e-coaching platform helping personal trainers guide athletes toward healthier lifestyles.
€211K funding raised 5× workload reduction 100K+ downloadsWorked within Eurecat's e-Health and Big Data Analytics units. Designed and developed Fairness and User Modeling algorithms to enhance personal trainer productivity — this work directly led to the first fairness-of-exposure algorithms for athletic ranking in the literature, later published as part of my doctoral research.
Designed, developed, and deployed user recommendation, face recognition, and sentiment analysis algorithms for the Shukran social network.
500K+ downloadsWeb and mobile development across several companies. Led development of CIH Bank's mobile banking app (1M+ downloads), contributed to Dubai Plus payments platform, worked on PMU Financial Information System, and co-founded a startup helping small businesses digitalize workflows.
1M+ downloads (CIH Bank)University of Cagliari, Italy · 2018–2021
Using ML to optimize personal trainer workload in e-coaching platforms. Focus: Recommender Systems, Algorithmic Fairness, Dropout Prediction, bias-aware rankings.
University of Fez, Morocco · 2012–2014
Software Engineering, Web Engineering, Design Patterns, Data Mining, Databases, Image Processing, Research Operations.
University of Rabat, Morocco · 2009–2012
Mathematics & Computer Science — Software Engineering option. Mathematical foundations, Programming, Data Structures, Algorithms.
First in literature to introduce fairness-of-exposure algorithms for athletic e-coaching. Models athlete performance and produces gender-fair urgency rankings for coaches. Validated on a real-world dataset.
Open-source LLM published on HuggingFace — a 3B-parameter instruction-tuned model with multilingual adaptations for enterprise AI and code generation contexts.
Trained a Convolutional Variational Autoencoder (CVAE) on Fashion MNIST to reconstruct fashion images and learn compact latent representations. Leveraged the latent space as a similarity index for visual search — conceptually analogous to semantic search in modern RAG architectures, but applied to image retrieval.
Core NLP engine for theShukran social network (500K+ downloads). Detects language, translates to English, and classifies sentiment using Naive Bayes + SentiWordNet + SenticNet.
Ensemble Deep Learning model classifying Scoliosis vs healthy subjects from non-invasive Rasterstereography analysis. Built with Keras multi-model fusion architecture.
Image classification using a large-scale dataset with small label noise. Trained a classifier to rank data for annotators — demonstrated state-of-the-art results with a fine-tuned deep CNN.
Whether you're exploring an AI partnership, interested in research collaboration, or looking to discuss manufacturing AI — I'm always open to meaningful conversations.