CV
PhD candidate in Artificial Intelligence (graduating February 2027) in the National PhD Program at Sapienza University of Rome, hosted at SisInfLab, Polytechnic University of Bari. I specialize in multimodal representation learning with LLMs and large vision–language models (LVLMs) for retrieval and recommendation: building generalist embeddings on top of them, auditing their benchmarks, and improving their robustness and efficiency (sparse vs. dense).
Work experience
Doctoral Researcher — Multimodal AI and Information Retrieval
- University: SisInfLab @ Polytechnic University of Bari, Italy
- Period: Nov 2023 - ongoing
- Description: Showed that generalist text embeddings beat specialized baselines for zero-shot recommendation and search (ACM RecSys 2025), and that LVLM-derived multimodal-by-design embeddings outperform fusion pipelines (ACM CIKM 2025). Working on post-hoc sparsification of frozen multimodal embeddings via sparse autoencoders for efficient retrieval. Co-developer and maintainer of Ducho 2.0 (TheWebConf 2024) and author of the first large-scale multimodal recommendation benchmark (ESWA 2025). I train, fine-tune (LoRA/PEFT), and evaluate LLMs and VLMs on multi-GPU HPC clusters using PyTorch DDP, Hugging Face Accelerate, Slurm, Kubernetes, and Weights & Biases.
Visiting PhD Student
- University: Alpen-Adria-Universität Klagenfurt, Austria
- Period: Jun 2026 - Aug 2026
- Description: Working in Prof. Dietmar Jannach’s group, exploring alternatives to classical BPR-style optimization for personalization with general-purpose multimodal representations and LVLMs.
Visiting PhD Student
- University: University of Edinburgh, School of Informatics, UK
- Period: Mar 2026 - May 2026
- Description: Audited composed image retrieval benchmarks with 12 multimodal embedding models in Dr. Pasquale Minervini’s group, exposing unimodal shortcuts and releasing human-validated splits and code. Also investigated limits of embedding-based retrieval in RAG (unargmaxability) and the robustness of generalist multimodal embeddings to query perturbations.
Assistant Lecturer — Foundations of Machine Learning
- University: Polytechnic University of Bari, Italy
- Period: Jul 2024 - ongoing
- Description: Design and deliver lab sessions, assignments, and lectures for the MSc Foundations of Machine Learning course.
Deep Learning Engineer, R&D
- Company: Wideverse, Italy
- Period: Apr 2022 - Sep 2023
- Description: Solved an industry partner’s real-time mobile perception use case end-to-end: trained YOLO-family detectors and deployed them on-device for soft real-time smartphone inference via TFLite (Android) and CoreML (iOS). Developed a visually-aware outfit recommender for a fashion industry partner, seeding later published research.
Research Intern
- University: INFSYS Lab @ Alpen-Adria-Universität Klagenfurt, Austria
- Period: Apr 2023 - Jun 2023
- Description: Developed compatibility-aware fashion recommendation for unseen item combinations using GANs, Stable Diffusion, and autoregressive models, supervised by Prof. Dietmar Jannach; basis of the GeCo journal paper (Information Sciences). Additionally, I completed my Master’s Thesis under the joint supervision of Prof. Dietmar Jannach and Prof. Tommaso Di Noia.
Education
Ph.D. in Artificial Intelligence
- University: Sapienza University of Rome — hosted at SisInfLab, Polytechnic University of Bari
- Period: Nov 2023 - Feb 2027 (expected)
- Description: National PhD Program in Artificial Intelligence. Supervisors: Prof. Tommaso Di Noia and Prof. Dietmar Jannach.
MSc. in Computer Engineering
- University: Polytechnic University of Bari
- Period: Oct 2021 - Oct 2023
- Grade: 110/110 with honors (full marks)
- Description: Thesis entitled “Generative AI for Complementary Item Recommendation in the Fashion Domain”.
BSc. in Computer Science and Automation Engineering
- University: Polytechnic University of Bari
- Period: Oct 2018 - Jul 2021
- Grade: 110/110 with honors (full marks)
- Description: Thesis entitled “Explainable AI: analysis of algorithms and application in the Recommender Systems area”.
Technical expertise
- Research: LLMs and large vision–language models (LVLMs); multimodal representation learning; generalist multimodal embeddings (dense and sparse); information retrieval and RAG; recommender systems; generative AI (diffusion, flow matching, GANs, VAEs); benchmark design and evaluation; robustness.
- ML frameworks: PyTorch, Hugging Face Transformers, Diffusers, Accelerate and PEFT (LoRA), Sentence Transformers, DSPY, vLLM, TensorFlow/Keras, scikit-learn, OpenCV, NumPy, Pandas.
- Retrieval and efficiency: FAISS, approximate nearest-neighbor search, sparse retrieval and inverted indexes, embedding compression and quantization.
- Systems and deployment: Slurm GPU clusters, Kubernetes, Docker, Weights & Biases, Linux, Spark, SQL; on-device inference with TFLite and CoreML.
- Programming: Python, Java, C.
Training
- Summer schools: 2nd Generative Modeling Summer School (GeMSS 2024), Eindhoven, Netherlands; ACM Europe School on Recommender Systems 2024.
Presentations
- FashionXRecSys Workshop at ACM RecSys 2022; Italian Information Retrieval Workshop 2024; ACM RecSys 2025 (one paper and one tutorial); ACM CIKM 2025.