About me.
Hi, I'm Riccardo Orlando, Computer Scientist and AI Researcher.
- based
- Rome, Italy
Contacts.
Now.
Senior AI Engineer – Principled Intelligence
Rome, Italy
Sept. 2026 –
Now
News.
September 2026
New Position at Principled Intelligence!
Rome, Italy
Started a new position as Senior AI Engineer at Principled Intelligence.
March 2025
New Position at DIME AI!
World Bank Group, Rome, Italy
Started a new position as AI Researcher at DIME AI.
December 2024
Minerva 7B Released!
Sapienza University of Rome, Italy
Public release of the Minerva 7B (base and instruct) model.
August 2024
ReLiK Code and Models released!
Sapienza University of Rome, Italy
Public release of the code and models for ReLiK, our system for Entity Linking and Relation Extraction.
May 2024
ReLiK paper accepted at ACL 2024!
Sapienza University of Rome, Italy
Our paper "ReLiK: Retrieve and LinK, Fast and Accurate Entity Linking and Relation Extraction on an Academic Budget" has been accepted at ACL 2024.
May 2024
Minerva LLM Family Released!
Sapienza University of Rome, Italy
Public release of the Minerva models (350M, 1B, 3B), the first family of Large Language Models (LLMs) trained from scratch on Italian (and English) data.
Publications.
Predicting Causal Effects from Natural Language Queries using Structured Representations
Giuliano Martinelli, Piriyakorn Piriyatamwong, Abelardo Carlos Martinez Lorenzo, Jasmin Baier, Riccardo Orlando, Satvik Garg, Sharif Kazemi, Linxi Wang, Arianna Legovini, and Samuel Fraiberger
arXiv preprint, 2026
May 2026
RAED: Retrieval-Augmented Entity Description Generation for Emerging Entity Linking and Disambiguation
Karim Ghonim, Pere-Lluís Huguet Cabot, Riccardo Orlando, and Roberto Navigli
In Proceedings of EMNLP, 2025
November 2025
ZEBRA: Zero-Shot Example-Based Retrieval Augmentation for Commonsense Question Answering
Francesco Maria Molfese, Simone Conia, Riccardo Orlando, and Roberto Navigli
In Proceedings of EMNLP, 2024
November 2024
Minerva LLMs: The First Family of Large Language Models Trained from Scratch on Italian Data
Riccardo Orlando, Luca Moroni, Pere-Lluís Huguet Cabot, Simone Conia, Edoardo Barba, Sergio Orlandini, Giuseppe Fiameni, and Roberto Navigli
In CLiC-it, 2024
September 2024
ReLiK: Retrieve and LinK, Fast and Accurate Entity Linking and Relation Extraction on an Academic Budget
Riccardo Orlando, Pere-Lluís Huguet Cabot, Edoardo Barba, and Roberto Navigli
In Findings of ACL, 2024
March 2024
MOSAICo: a Multilingual Open-text Semantically Annotated Interlinked Corpus
Simone Conia, Edoardo Barba, Abelardo Carlos Martinez Lorenzo, Pere-Lluís Huguet Cabot, Riccardo Orlando, Luigi Procopio, and Roberto Navigli
In Proceedings of NAACL, 2024
June 2024
Exploring Non-Verbal Predicates in Semantic Role Labeling: Challenges and Opportunities
Riccardo Orlando, Simone Conia, and Roberto Navigli
In Findings of ACL, 2023
July 2023
Universal Semantic Annotator
Roberto Navigli, Riccardo Orlando, Cesare Campagnan, and Simone Conia
In European Language Grid. Springer, Cham, 2023
November 2022
Universal Semantic Annotator: the First Unified API for WSD, SRL and Semantic Parsing.
Riccardo Orlando, Simone Conia, Stefano Faralli, and Roberto Navigli
In Proceedings of LREC, 2022
June 2022
AMuSE-WSD: An All-in-one Multilingual System for Easy Word Sense Disambiguation.
Riccardo Orlando, Simone Conia, Fabrizio Brignone, Francesco Cecconi, and Roberto Navigli
In Proceedings of EMNLP, 2021
November 2021
InVeRo-XL: Making Cross-Lingual Semantic Role Labeling Accessible with Intelligible Verbs and Roles.
Simone Conia, Riccardo Orlando, Fabrizio Brignone, Francesco Cecconi, and Roberto Navigli
In Proceedings of EMNLP, 2021
November 2021
Past.
Researcher – World Bank Group
Rome, Italy
Built ZeroHungerAI's LLM extraction system from the ground up: training data, models, evaluation benchmark, and the inference setup that runs it across 150 million news articles. Also built the retrieval and agentic parts of ImpactAI's meta-analysis pipeline, plus the RCT benchmark used to evaluate retrieval, and got it into production.
Mar. 2025 –
Aug. 2026
PhD – Sapienza NLP Group
Sapienza University of Rome, Italy
PhD in Natural Language Processing and Deep Learning at the Sapienza NLP group of Sapienza University of Rome.
Nov. 2021 –
Jan. 2025
Feb. 2021 –
Dec. 2021
Master of Science in Engineering in Computer Science
Sapienza University of Rome, Italy
Thesis title: An automatic approach to produce multilingual training data for Semantic Role Labelling.
Grade: 110/110 with honours
2018 –
2020
Software Engineer – Capgemini
Rome, Italy
Included in a team that worked to re-platform a series of internal applications of an insurance company. Worked as a software engineer mainly on the backend but also contributed to the frontend.
Mar. 2018 –
Sept. 2018
Bachelor of Computer Science
Sapienza University of Rome, Italy
Thesis title: Densest subgraph in Fork/Join.
2015 –
2018
Personal projects.
These are some projects I'm involved in:
🦮 Golden Retriever
A Lightning framework for retriever architecture prototype
🦙 Llama Trainer Utility
A "just few lines of code" utility for fine-tuning (not only) Llama models.
transformers-embedder
A Word Level Transformer layer based on Pytorch and 🤗Transformers.
nlp-template
A template to initialize PyTorch projects that use as a backbone framework PyTorch Lightning.
🍺 IPA: import, preprocess, accelerate
NLP Preprocessing Pipeline Wrappers.
ner-serve
Simple multilingual NER model serving using Docker, FastAPI, ONNX and Multilingual Mini-LM.
transformers-ner
Simple NER model, showcasing Transformer Embedder library.
transformer-srl
Reimplementation of a BERT based model (Shi et al, 2019), currently the state-of-the-art for English SRL. This model implements also predicate disambiguation.
Super SloMo TF2
Tensorflow 2 implementation of Super Slo Mo paper.
Chinese word segmentation
Chinese Word Segmentation task based on BERT and implemented in Pytorch for state-of-the-art Chinese word segmentation.
Word Sense Disambiguation
A supervised Bi-LSTM architecture for the Word Sense Disambiguation task.
Sense Embeddings
BabelNet (and WordNet) sense embedding trained with Word2Vec and FastText.
CNN Image Classification
Convolutional Neural Networks trained to classify different types of boats.
Machine Learning Malware Analysis
Three different classifier trained to distinguish malware application from non-malware ones and recognize the family they belong.
Densest Subgraph in Fork/Join
Bachelor’s thesis. Densest Subgraph in Fork/Join. A Fork/Join parallel algorithm for the densest subgraph problem.
Fork/Join Sudoku Solver
Solver that uses Fork/Join framework to solve Sudoku in parallel, made during a multi-core programming course.