Marco Trombetti

Last updated
Marco Trombetti
Born (1976-03-14) March 14, 1976 (age 48)
Rome, Italy
NationalityItalian
Occupations
  • Computer scientist
  • entrepreneur
  • investor
Known for Translated
Matecat
ModernMT
Pi Campus
Pi School
Translated 9
SpouseIsabelle Andrieu
Website marcotrombetti.com

Marco Trombetti (born March 14th, 1976) is an Italian computer scientist, entrepreneur, investor, and ocean sailor. He co-founded Translated, a pioneer of artificial intelligence in the language industry. Through Translated, he helped develop the first AI-powered open-source Computer Assisted Translation (CAT) tool, Matecat, [1] which also introduced the first adaptive machine translation system. [2] He is considered one of the most influential leaders and innovators in the language industry. [3] His research on progress toward the language singularity, [4] [5] presented during a keynote at the Association for Machine Translation in the Americas (AMTA) conference in 2022, [6] has provided invaluable insights into the field of artificial intelligence. [7] [8] [9] [10]

Contents

Early life and education

Trombetti was raised in Rome and started programming at a very early age. Between 14 and 19, Trombetti distributed software he developed on Fidonet BBS. Among the software he distributed was ArcyCD, which efficiently converted music CDs into cassettes, and MoneyMaker, which allowed small shop owners to use barcode readers, which were only available to the largest store. While studying astrophysics at Roma Tre University, Trombetti continued developing technical solutions, specifically for the digital mapping, information retrieval, and image manipulation markets. In 1999, with Gianluca Granero, he created an online community WebChatWorld, whose traffic asset was acquired by Doubleclick, now part of Google; the capital gained was used to fund Translated.

Business career

Translated

During an Erasmus Programme year at the Joseph Fourier University in Grenoble (France), Trombetti met his future co-founder and wife, Isabelle Andrieu. In 1999, he dropped out of Roma Tre University, and together, they founded Translated, one of the first online translation services. They aimed to “allow everyone to understand and be understood in their own language” and promote universal communication. By providing translators with the most advanced translation software free of charge, they aimed to foster a mutually beneficial relationship between translators and machine translation, where translators would teach machines to translate better, and machines would improve in real time to provide better translation suggestions. To realize this vision as CEO and CTO of Translated, Trombetti led the investment in research and development of a computer-assisted translation (CAT) tool to speed up the translation process and the first adaptive machine translation system to enable mutually successful human-machine interaction. In 2022, Gartner [11] recognized Translated as a leading provider of AI-enabled translation services. The following year, the IDC MarketScape [12] acknowledged Translated's Adaptive MT system as the world's leading enterprise translation AI.

Memopal

In 2007, Trombetti, together with Gianluca Granero, started Memopal, one of the first cloud storage companies. Marco served the company as the CEO until the company was acquired by Defenx, [13] a publicly listed security company.

Pi Campus

In 2007, Trombetti, along with Isabelle Andrieu and Gianluca Granero, founded Pi Campus, first as a place to host the investments he made in startups, then as a startup district and later as a venture firm. In 2016, former CVC Capital Partners co-founder Roberto Lombardi joined the company by acquiring an undisclosed amount of shares. Trombetti oversees Pi Campus' investments, which include more than 50 in Europe and the U.S. It is one of the many firms that regularly invests with Y Combinator in startups that have just graduated from Y Combinator's accelerator program. Under the Pi Campus brand, Trombetti published a book, The New Prince, exploring the counterintuitive reasons and strategies that have led new generation of entrepreneurs to success. The book is a collection of essays written by Trombetti to answer the recurring questions asked by founders of the startups he has invested in.

ModernMT

In 2017, together with researcher Marcello Federico and funded by Translated and the Fondazione Bruno Kessler, Trombetti led the creation of ModernMT, the first commercial application of the Transformer (deep learning architecture) technology to provide adaptive neural machine translation. In 2023, an independent evaluation by CSA Research [14] recognized ModernMT as the most advanced implementation of responsive MT. The same year, researcher and MT consultant Achim Ruoop designed an MT evaluation and comparison toolkit to compare adaptive and generic MT solutions in a real-world scenario and test ModernMT against other leading, generic, publicly available solutions. The results confirmed ModernMT's ability to perform better than any competitor. In 2024, Translated repeated the test with the latest version of the MT systems tested by Ruoop and added GPT-4 to the competition. ModernMT performed the best in most evaluation tests. [15]

Pi School

In 2017, together with Jamshid Alamuti, Trombetti founded Pi School, an innovative educational venture to develop a new class of AI specialists. Pi School provides personalized coaching from world-renowned AI authorities and experienced engineers selected through a global application process. Students must apply their new skills to industry projects provided by their employers, tech companies, corporations, or fast-growing startups. Pi School's scientific advisory board includes Marcello Federico (senior principal scientist at Amazon), Hassan Sawaf (Founder of aiXplain and former AI Director at Facebook and Amazon), and Alexander Waibel (former director of language technologies at Facebook).

Contributions to the language industry

Open-source CAT tool with built-in machine translation

In 2010, Trombetti began promoting the idea of a free CAT tool with a built-in neural machine translation system that could learn from corrections in real time. He rallied the Fondazione Bruno Kessler, the Le Mans University, and the University of Edinburgh around the project. The newly formed consortium won a three-year grant from the European Union's Seventh Framework Program for Research, Technological Development and Demonstration. [16] The consortium set aside the neural machine translation component and focused on the tool, released as open-source software in 2014 under the name MateCat. The first version of Matecat included an adaptive static MT system developed by Translated, allowing the user to use a different machine translation solution. Trombetti was involved in the design of the tool and led the development team, pursuing research in machine learning models that could integrate machine translation into a human translation workflow. In 2015, Matecat won the TAUS Game Changer Innovation Contest for the development of an integrated, adaptive, static machine translation system inside a CAT tool. [17] The same year, it was recognized as one of the most successful AI initiatives funded by the 7th Research Framework Programme of the European Union. [18]

Adaptive neural machine translation

In 2014, as a follow-up of the Matecat project, Trombetti organized a new research consortium consisting of Translated, Fondazione Bruno Kessler, the University of Edinburgh, and TAUS to develop a neural machine translation system that could improve in real-time based on the corrections it received. [19]

In 2017, the resulting system, backed by a research grant from the European Union through the 2020 Innovation Action (2015–2017), debuted as ModernMT. [20]

Ocean sailing

In 2020, companies still considered the adoption of AI-based translation solutions risky. [21] Recognizing that the spray-and-pray approach wasn't working to sell Translated's new AI-first localization platform, TranslationOS, [22] Trombetti began looking for a smart way to identify innovators and early adopters in the translation industry. Despite being completely new to ocean sailing, he decided to participate in the Ocean Globe Race, a round-the-world regatta celebrating the 50th anniversary of the Whitbread Round the World Race, which limited the use of technology and the presence of professionals on board. Translated's sailboat was named Translated 9, a reference to T9 (predictive text) technology, the first language model used to assist typing on early cell phones. Trombetti called out for interested parties worldwide to assemble the crew, including himself as co-skipper. A twin boat in San Francisco was dedicated to providing an offshore training experience [23] for people in the localization industry who wondered if they could become part of the Translated 9's crew. The training in San Francisco was conducted by the famous yachtsman Paul Cayard, winner of the 1998 Whitbread. This approach allowed Translated to understand potential customers' propensity for adventure and, therefore, for risk and innovation.

In preparation for the Ocean Globe Race, Trombetti participated in the 2023 Cape 2 Rio Race with Translated 9. This was his first ocean regatta.

On the hull of Translated 9, Trombetti wrote: "We believe in humans." This motto expresses the company's belief in people who challenge themselves to go beyond what others think is possible. Translated 9 won legs 1 and 2 of the Ocean Globe Race 2023 and was leading leg 3 until it was forced to retire after successfully rounding Cape Horn due to a hull breach. Under Trombetti's leadership, [24] the crew was able to repair the hull and return for the start of Leg 4 to complete the round-the-world regatta. In March 2024, SeaHorse Magazine named Trombetti Sailor of the Month. [25] In May 2024, Il Giornale della Vela named Trombetti boat owner of the Year. [26]

Personal life

Trombetti is married to linguist Isabelle Andrieu and is the father of three children. He is fluent in Italian, English, and French.

See also

Related Research Articles

<span class="mw-page-title-main">Machine translation</span> Computerized translation between natural languages

Machine translation is use of computational techniques to translate text or speech from one language to another, including the contextual, idiomatic and pragmatic nuances of both languages.

Computer-aided translation (CAT), also referred to as computer-assisted translation or computer-aided human translation (CAHT), is the use of software, also known as a translator, to assist a human translator in the translation process. The translation is created by a human, and certain aspects of the process are facilitated by software; this is in contrast with machine translation (MT), in which the translation is created by a computer, optionally with some human intervention.

In the history of artificial intelligence, an AI winter is a period of reduced funding and interest in artificial intelligence research. The field has experienced several hype cycles, followed by disappointment and criticism, followed by funding cuts, followed by renewed interest years or even decades later.

Kaltura, Inc. is a New York-based software company founded in 2006. It operates in several major markets: webinars and virtual events, enterprise video content management and online video platform (OVP), educational technology, and Cloud TV software, and offers products such as video portal, LMS and CMS extension, virtual event and webinar platform, and TV streaming app.

<span class="mw-page-title-main">Box, Inc.</span> Cloud content management program

Box, Inc. is a public company based in Redwood City, California. It develops and markets cloud-based content management, collaboration, and file sharing tools for businesses. Box was founded in 2005 by Aaron Levie and Dylan Smith. Initially, it focused on consumers, but around 2009 and 2010 Box pivoted to focus on business users. The company raised about $500 million over numerous funding rounds before going public in 2015. Its software allows users to store and manage files in an online folder system accessible from any device. Users can then comment on the files, share them, apply workflows, and implement security and governance policies.

<span class="mw-page-title-main">Figure Eight Inc.</span> American software company

Figure Eight was a human-in-the-loop machine learning and artificial intelligence company based in San Francisco.

Arabic is one of the major languages that have been given attention by machine translation (MT) researchers since the very early days of MT and specifically in the U.S. The language has always been considered "due to its morphological, syntactic, phonetic and phonological properties [to be] one of the most difficult languages for written and spoken language processing."

ProMT is a lead Russian developer of language translation software for businesses and private users since 1991. The company provides on-premises software based on neural technologies.

<span class="mw-page-title-main">Jeff Dean</span> American computer scientist and software engineer

Jeffrey Adgate "Jeff" Dean is an American computer scientist and software engineer. Since 2018, he has been the lead of Google AI. He was appointed Google's chief scientist in 2023 after the merger of DeepMind and Google Brain into Google DeepMind.

Google Brain was a deep learning artificial intelligence research team that served as the sole AI branch of Google before being incorporated under the newer umbrella of Google AI, a research division at Google dedicated to artificial intelligence. Formed in 2011, it combined open-ended machine learning research with information systems and large-scale computing resources. It created tools such as TensorFlow, which allow neural networks to be used by the public, and multiple internal AI research projects, and aimed to create research opportunities in machine learning and natural language processing. It was merged into former Google sister company DeepMind to form Google DeepMind in April 2023.

OpenNN is a software library written in the C++ programming language which implements neural networks, a main area of deep learning research. The library is open-source, licensed under the GNU Lesser General Public License.

Feedzai is a data science company that develops real-time machine learning tools to identify fraudulent payment transactions and minimize risk in the financial services, retail, and e-commerce industries. The company has been classified as a unicorn startup since March 2021, after a Series D funding round pushed its value above $1 billion.

<span class="mw-page-title-main">Paxata</span> American private software company

Paxata is a privately owned software company headquartered in Redwood City, California. It develops self-service data preparation software that gets data ready for data analytics software. Paxata's software is intended for business analysts, as opposed to technical staff. It is used to combine data from different sources, then check it for data quality issues, such as duplicates and outliers. Algorithms and machine learning automate certain aspects of data preparation and users work with the software through a user-interface similar to Excel spreadsheets.

MateCat is a web-based computer-assisted translation (CAT) tool, released as open-source software under the Lesser General Public License (LGPL).

Robotic process automation (RPA) is a form of business process automation that is based on software robots (bots) or artificial intelligence (AI) agents. RPA should not be confused with artificial intelligence as it is based on automotive technology following a predefined workflow. It is sometimes referred to as software robotics.

Google Neural Machine Translation (GNMT) was a neural machine translation (NMT) system developed by Google and introduced in November 2016 that used an artificial neural network to increase fluency and accuracy in Google Translate. The neural network consisted of two main blocks, an encoder and a decoder, both of LSTM architecture with 8 1024-wide layers each and a simple 1-layer 1024-wide feedforward attention mechanism connecting them. The total number of parameters has been variously described as over 160 million, approximately 210 million, 278 million or 380 million. It used WordPiece tokenizer, and beam search decoding strategy. It ran on Tensor Processing Units.

The Open Neural Network Exchange (ONNX) [] is an open-source artificial intelligence ecosystem of technology companies and research organizations that establish open standards for representing machine learning algorithms and software tools to promote innovation and collaboration in the AI sector. ONNX is available on GitHub.

Clarifai is a leader in AI orchestration and development, helping organizations, teams, and developers build, deploy, orchestrate, and operationalize AI at scale. Clarifai’s cutting-edge AI workflow orchestration platform leverages today's modern AI technologies like Large Language Model(LLMs), Large Vision Model (LVMs), and Retrieval Augmented Generation (RAG), data labeling, inference, and more, and is available in cloud, on-premises, or hybrid environments. Founded in 2013, Clarifai has been used to build more than 1.5 million AI models with more than 400,000 users in 170 countries.

<span class="mw-page-title-main">Multiverse Computing</span> Quantum computing company

Multiverse Computing is a Spanish quantum computing software company headquartered in San Sebastián, Spain, with offices in Paris, Munich, London, Toronto and Sherbrooke, Canada. The Spanish startup applies quantum and quantum-inspired algorithms to problems in energy, logistics, manufacturing, mobility, life sciences, finance, cybersecurity, chemistry, materials science and aerospace.

Open-source artificial intelligence is an AI system that is freely available to use, study, modify, and share. These attributes extend to each of the system's components, including datasets, code, and model parameters, promoting a collaborative and transparent approach to AI development. Free and open-source software (FOSS) licenses, such as the Apache License, MIT License, and GNU General Public License, outline the terms under which open-source artificial intelligence can be accessed, modified, and redistributed.

References

  1. ↑ The Matecat Tool
  2. ↑ MateCat: an Open Source CAT Tool for MT post-­‐edi8ng
  3. ↑ Multilingual Magazine, February 2024
  4. ↑ This Is the Speed at Which We Are Approaching Singularity in AI
  5. ↑ Measuring User Productivity in Machine Translation Enhanced Computer Assisted Translation
  6. ↑ AMTA 2022 Orlando
  7. ↑ Multilingual Magazine, February 2021
  8. ↑ The Next Web, January 2023
  9. ↑ Popular Mechanics Magazine, June 2023
  10. ↑ UniCamillus Global Health Journal
  11. ↑ "Translated named in the 2022 Gartner® Market Guide for AI-Enabled Translation Services".
  12. ↑ Machine Translation Software 2022 Vendor Assessment
  13. ↑ Defenx acquires Memopal
  14. ↑ ModernMT and the Promise of Responsive MT
  15. ↑ New Research Confirms ModernMT Outperforms Leading MT and GenAI Solutions for Enterprise Translations
  16. ↑ Machine Translation Enhanced Computer Assisted Translation
  17. ↑ "Matecat Wins Taus Game Changers Innovation Contest in Shenzhen | TAUS". Archived from the original on 2017-07-21.
  18. ↑ Two Translation Technology Companies Nominated for New EU innovation Radar Award
  19. ↑ MMT: New open source MT for the translation industry
  20. ↑ "Modernmt/Modernmt". GitHub .
  21. ↑ Gartner Market Guide for AI-Enabled Translation Services
  22. ↑ Translated's TranslationOS
  23. ↑ "A Daysail with Paul Cayard and the Ocean Globe Race". 13 May 2022.
  24. ↑ Applying the Startup Approach to Sailing Around the World
  25. ↑ "Seahorse Sailor of the Month Hall of Fame".
  26. ↑ "Velista dell'anno (Giornale della Vela), da Rossetti a Trombetti tutti i vincitori". 26 May 2024.
This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.