Piero Molino

Career

Piero Molino co-founded QuestionCube, a startup working on question answering, and served as its CTO from 2011 to 2015; Molino's PhD thesis, completed in 2015, was on semantics and question answering. Molino then worked as a researcher at IBM Watson on question understanding, cognitive search and text classification, and in 2016 as a machine learning scientist at Geometric Intelligence.¹

After Uber acquired Geometric Intelligence and created Uber AI Labs, Molino worked there from 2017 to 2020 on natural language processing, dialogue systems and graph learning, rising to Senior Research Scientist II. Molino spent 2020 to 2021 as a Staff Research Scientist in Chris Ré's Hazy Research group at Stanford, co-founded Predibase in 2020 with Travis Addair, Dev Rishi and Chris Ré, and in 2025 co-founded Studio Atelico, which Molino leads as CEO.¹

Open-source work

At Uber AI Labs Molino created Ludwig, a declarative deep learning framework described in a 2019 paper with Yaroslav Dudin and Sai Sumanth Miryala. The Apache-2.0 project is now hosted by the Linux Foundation AI & Data. At Predibase, Ludwig and the LoRAX multi-LoRA inference server formed the basis for fine-tuning and serving language models.¹³²⁴

Research

Molino's publications include the CoordConv paper on a failing of convolutional networks (NeurIPS 2018), Plug and Play Language Models for controlled text generation (ICLR 2020), and a 2021 NeurIPS paper on personalized benchmarking with the Ludwig Benchmarking Toolkit. Other projects listed on Molino's site include the Plato conversational AI platform, the COTA customer support assistant and the Manifold model visualization tool.⁵⁶⁷⁸

Work and contributions

• Ludwig Open-source code

  Molino's site says Molino created Ludwig, a declarative deep learning framework, at Uber AI Labs. The Apache-2.0 repository says Ludwig is hosted by the Linux Foundation AI & Data.¹²

• LoRAX Open-source code

  Molino's CV says Predibase made fine-tuning and serving LLMs easier through Ludwig and LoRAX; the Apache-2.0 LoRAX repository describes it as a multi-LoRA inference server.¹⁴

• Plug and Play Language Models (PPLM) Published research

  Co-author of the 2019 paper presenting a simple approach to controlled text generation, published at ICLR 2020.⁷⁵

Timeline

1. 2011Co-founded the question answering startup QuestionCube and served as its CTO.¹
2. 2015Joined IBM Watson as a researcher on NLP and question answering.¹
3. 2016Worked as a machine learning scientist at Geometric Intelligence.¹
4. 2017Became a research scientist at the newly created Uber AI Labs.¹
5. Jul 9, 2018Released the CoordConv paper on convolutional neural networks with Rosanne Liu, Jason Yosinski and other co-authors.⁶
6. Sep 17, 2019Released the paper describing Ludwig, a type-based declarative deep learning toolbox.³
7. Dec 4, 2019Released Plug and Play Language Models as a co-author.⁷
8. 2020Co-founded Predibase and joined Stanford's Hazy Research group as a Staff Research Scientist.¹
9. 2025Co-founded Studio Atelico, an AI-first games studio, as CEO.¹

Other documented roles

Roles the sources document that are not confirmed as current.

• CEO, CSO and Co-founder (2020–2025), Predibase¹
• Staff Research Scientist, Hazy Research group (2020–2021), Stanford University¹
• Research Scientist to Senior Research Scientist II (2017–2020), Uber AI Labs¹
• Machine Learning Scientist (2016), Geometric Intelligence¹
• Researcher (2015–2016), IBM Watson¹
• CTO and Co-founder (2011–2015), QuestionCube¹

Where they work

• Studio Atelico 1 person in the wiki

Related people

• Dong YuML frameworks and libraries · Language models
• Thomas J. FanML frameworks and libraries · Developer tools
• Abubakar AbidDeveloper tools · ML frameworks and libraries
• Simon WillisonLanguage models · Developer tools
• James BergstraML frameworks and libraries · Developer tools
• Craig FederighiLanguage models · Developer tools
• Jonathan CohenLanguage models · ML frameworks and libraries
• Matthew WatsonML frameworks and libraries · Developer tools
• David TalbyLanguage models · ML frameworks and libraries
• Clément FarabetLanguage models · ML frameworks and libraries