Siva Reddy
Siva Reddy is a computer scientist who develops AI methods for language understanding and reasoning. He pursued a PhD at the University of Edinburgh with a Google PhD Fellowship in 2015, completed a postdoctoral role at Stanford's NLP Group by September 2019, and by January 2020 was a joint faculty member at McGill and a Facebook CIFAR AI Chair at Mila. His work includes CoQA, StereoSet, MaRVL, StarCoder, and LLM2Vec, spanning conversational question answering, stereotype-bias measurement, cross-cultural visual reasoning, code models, and text embeddings. He received the CS-CAN Outstanding Early Career Computer Science Researcher Award in 2025 and a Sloan Research Fellowship in 2026. As of October 2026, he is an AI Scientist at Mistral and an Associate Professor on leave at McGill, a Canada CIFAR AI Chair, and a core academic member of Mila.
- Now at
- Mistral AI, McGill University and Mila
- Works
- 249
- Citations
- 6,364
- h-index
- 34
Publication numbers from OpenAlex, as of 27 October 2026.
Timeline
-
2015
A Google PhD Fellowship
While pursuing his PhD at the University of Edinburgh, he receives support from the Google PhD Fellowship in 2015.
-
By 2019
Postdoctoral work at Stanford
At Stanford University's NLP Group, he works as a postdoctoral researcher until September 2019.
-
2019
CoQA for conversational question answering
He coauthors CoQA in 2019, a dataset with 127,000 questions and answers from 8,000 conversations about text passages.
-
By January 2020
Joint faculty member at McGill
By January 2020, he is a joint faculty member in McGill’s School of Computer Science and Department of Linguistics and holds a Facebook CIFAR AI Chair at Mila. His research builds symbolic and deep-learning models for language understanding, including question answering and conversational systems.
-
2020
StereoSet and a VentureBeat award
He wins the VentureBeat AI Innovation Award in NLP/NLU in 2020 for his work on StereoSet, a dataset that measures stereotype bias in language models.
-
2021
MaRVL and cross-cultural reasoning
He coauthors MaRVL, a multilingual visual-reasoning dataset built from native-speaker annotations in Indonesian, Mandarin, Swahili, Tamil, and Turkish. The paper wins the Best Long Paper award at EMNLP in 2021.
-
2023–2024
StarCoder and LLM2Vec
In 2023, he coauthors StarCoder, a code model from the BigCode open-science collaboration; StarCoderBase is trained on one trillion tokens from The Stack. In 2024, he coauthors LLM2Vec, an unsupervised method that transforms decoder-only language models into text encoders.
-
2025
A CS-CAN early-career award
He receives the CS-CAN Outstanding Early Career Computer Science Researcher Award in 2025.
-
2026
An Alfred P. Sloan Fellowship
He receives the Alfred P. Sloan Research Fellowship in 2026.
-
As of October 2026
Mistral and McGill roles
As of October 2026, he is an AI Scientist at Mistral and its first researcher in Montreal. He is an Associate Professor in McGill's School of Computer Science and Department of Linguistics, currently on leave, and a Canada CIFAR AI Chair; he also co-leads the McGill NLP Group. He is a core academic member of Mila.
Career
- NowMistral AIAI Scientist
- NowMcGill UniversityAssociate Professor, on leave
- NowMcGill UniversityCo-lead, McGill NLP Group
- NowMilaCore academic member
- Until 2019Stanford UniversityPostdoctoral researcher, NLP Group
- McGill UniversityJoint faculty member
- MilaFacebook CIFAR AI Chair
Education
Sources and credits
Photo: We have no published photo of Siva Reddy yet. Send us a free-licensed photo.
- The School of Computer Science Welcomes Four New Faculty Members
- Reasoning and Learning Lab | Siva Reddy
- CoQA: A Conversational Question Answering Challenge
- Siva Reddy Honored for Outstanding Early Career Computer Science Research | Mila
- Prof. Siva Reddy won the VentureBeat AI Innovation Award 2020
- Visually Grounded Reasoning across Languages and Cultures
- Best Paper Awards
- StarCoder: may the source be with you!
- LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders
- Siva Reddy Named 2026 Sloan Research Fellow
- Intelligence artificielle Mistral ML4 : l’IA « libre » qui veut faire plus avec moins
- Welcome | Siva Reddy
- Siva Reddy | Mila