Hi, I am Massimo!
I am a Staff Research Engineer at Google DeepMind based in Zurich focusing on the multimodal understanding capabilities of large language models (including Gemini). I am also interested in multilinguality, especially on synthetic data creation for resource-scarce languages. I got my PhD from the University of Trento where I worked on structural kernels and deep learning methods for Question Answering (QA).
Check out my latest papers on my Google Scholar profile or read more about me here.
News
- [Jul 2024] Our paper Translation and Transliteration Based Data Augmentation for Multilingual Semantic Parsing has been accepted at ECAI 2024. See you in Santiago de Compostela in October!
- [May 2024] The new Gemini 1.5 tech report is out. Proud to have contributed to improving its multimodal capabilities!
- [Apr 2024] I will give a talk titled “Decoding visual data with GenAI” at the first GenAI Zurich conference on May the 29th.
- [Mar 2024] I will be at NAACL 2024 in June, see you in Mexico City!
- [Oct 2023] Our paper and benchmark XTREME-UP: A User-Centric Scarce-Data Benchmark for Under-Represented Languages has been accepted at EMNLP 2023. See you in Singapore!
- [Oct 2023] Our paper mmT5: Modular Multilingual Pre-Training Solves Source Language Hallucinations has been accepted at EMNLP 2023.
- [Aug 2023] I will pitch a project for the Google Outreach and Mentorship Programme at Deep Learning Indaba. See you in Accra!
- [May 2023] Happy to have won an Assistant Tech Impact Award!
- [May 2023] Our paper and benchmark XTREME-UP: A User-Centric Scarce-Data Benchmark for Under-Represented Languages is available at https://github.com/google-research/xtreme-up
- [May 2023] Our paper mmT5: Modular Multilingual Pre-Training Solves Source Language Hallucinations is available on arXiv
- [Dec 2022] We presented our winning approach for the Zero-Shot MMNLU-22 Semantic Parsing Challenge at the MMNLU-22 workshop colocated with EMNLP in Abu Dhabi. You can find all the details in our workshop paper: Evaluating Byte and Wordpiece Level Models for Massively Multilingual Semantic Parsing
- [Nov 2022] Exciting update! I am now a Staff Software Engineer in Research
- [Aug 2022] Our FabT5 team won the Zero-Shot MMNLU-22 Semantic Parsing Challenge organized by Amazon. Our winning submission was based on ByT5 and the synthetic data augmentation technique we published at EMNLP 2021: Translate & Fill. See you at EMNLP 2022 in Abu Dhabi!
- [Sep 2021] Our EMNLP 2021 paper Translate & Fill: Improving Zero-Shot Multilingual Semantic Parsing with Synthetic Data is now on arXiv
- [Aug 2021] Our paper Translate & Fill: Improving Zero-Shot Multilingual Semantic Parsing with Synthetic Data, has been accepted at EMNLP 2021 in Punta Cana
- [Nov 2020] I now hold the role of Senior Software Engineer in Research
- [Sep 2019] Our EMNLP 2019 paper Answering Conversational Questions on Structured Data without Logical Forms is now on arXiv
- [Aug 2019] I will attend the Conversational Search and Recommendation workshop at Google London
- [Aug 2019] Our Google Research paper, Answering Conversational Questions on Structured Data without Logical Forms, has been accepted at EMNLP 2019 in Hong Kong
- [Jul 2019] I will attend ACL 2019 in Florence. See you there!
- [Aug 2018] Semantic Linking in Convolutional Neural Networks for Answer Sentence Selection short paper accepted at EMNLP 2018. See you in Brussels!
- [Apr 2018] This summer I will join Google Zurich as a Software Engineer in Research
- [Apr 2018] I received my Ph.D. from the University of Trento!
- [Mar 2018] My colleague Antonio Uva and I won the best poster prize awarded by Seac during the ICT Days 2018 (the poster included my CIKM 2017 work)
- [Sep 2017] Received an honourable mention and travel award for my top ten solution at the CIKM AnalytiCup 2017: Lazada Product Title Quality Challenge
- [Aug 2017] Accurate Sentence Matching with Hybrid Siamese Networks short paper accepted at CIKM 2017
- [Aug 2017] I am a semi-finalist (top 10) at the CIKM 2017 Lazada Product Title Quality Challenge
- [Aug 2017] I have been invited to the Google Natural Language Processing Summit in Zurich (25-27 September)
- [Jun 2017] Learning Contextual Embeddings for Structural Semantic Similarity using Categorical Information long paper accepted at CoNLL 2017
- [Apr 2017] RelTextRank: An Open Source Framework for Building Relational Syntactic-Semantic Text Pair Representations demo paper accepted at ACL 2017
- [Feb 2017] Question Answering and Knowledge Graphs chapter published for Springer
- [Dec 2016] This summer I will be at Google Zurich to work on adversarial training for text generation