“Aiming not to leave any language behind”
- Marzieh Fadaee, AI4Her Legacy Award Finalist, on advancing inclusive multilingual AI
Marzieh Fadaee is the Head of Cohere Labs, where she guides research on multilingual language models, data efficiency, and evaluation. As a nominee for the AI4Her Legacy Award, she is acknowledged for her lasting contributions to multilingual AI, her leadership in open and responsible research, and her dedication to ensuring that language technologies serve diverse languages and communities.
Written by Myrthe van den Berg
With a PhD from the University of Amsterdam, Marzieh has spent over a decade shaping how the field approaches multilinguality, evaluation, and data-efficient learning. Before joining Cohere Labs, she led the research team at Zeta Alpha. Today, her work focuses on developing models with a deeper understanding of human language, especially for languages that have historically been excluded from AI development.
Her fascination with language started early. “In hindsight, it all began when I was 13 and discovered Tolkien,” Marzieh shares. “I was instantly captivated by the incredible world-building through the creation of various distinct languages.”
That early curiosity evolved into a research career focused on ensuring AI systems perform reliably across linguistic and cultural boundaries.
“We launched a year-long, multi-institutional effort to develop a multilingual language model, with the goal of including every language.”
Marzieh is currently involved in multiple research projects at the crossroads of AI and language understanding. One of her main efforts is a year-long, multi-institutional collaboration to develop a multilingual language model designed to assist speakers of many languages, not just a few.
“We launched a year-long, multi-institutional effort to develop a multilingual language model, with the goal of including every language,” she explains. “I’m leading the data team in this effort to gather and enhance data from volunteers around the world.”
Meanwhile, Marzieh’s work focuses on responsible AI development at a systems level. She helps identify safety and bias challenges, plans responsible open-source model releases, and participates in policy discussions about responsible AI use.
“My work on responsible AI focuses on providing practical guidelines and raising awareness about safe scenarios for model use, and what precautions are necessary to prevent toxic, biased, or harmful output,” she says.
This combination of technical rigor and ethical foresight has made Marzieh a leading figure in promoting trustworthy large language models. Her research on evaluation, efficiency, and data-aware learning, published in top venues such as ACL, EMNLP, ICLR, and NeurIPS, has helped establish new standards for how multilingual AI systems are evaluated and implemented.
Mentorship, Representation, and the Next Generation of AI Researchers
“Don’t wait for permission or be discouraged by others. Embrace that spark of passion because, just like any field, AI can benefit from diverse voices and perspectives.
Apart from her research contributions, Marzieh is deeply dedicated to mentorship and community building. She actively participates in various mentorship programs, including the ACL mentorship program, Deep Learning Indaba mentorship, WI mentorship, and personally supervises interns each year at her company.
She considers mentorship vital for the long-term health of the field, especially for newcomers starting in AI.
“If you are drawn to a small aspect of AI, don’t hesitate to pursue it,” she says. “Don’t wait for permission or be discouraged by others. Embrace that spark of passion because, just like any field, AI can benefit from diverse voices and perspectives. The three competencies you need are perseverance, adaptability, and nonconformity.”
She also highlights the importance of building strong foundations and learning from others. “Classic papers and books are important for setting the foundation of knowledge,” she notes. “Many open and public sources are available, so find what’s relevant for you and dig in.”
As an AI4Her Legacy Award nominee, Marzieh Fadaee exemplifies the values the award celebrates: lasting impact, inclusive innovation, and leadership that influences the field over time. Through her work on multilingual AI, responsible research, and mentorship, she continues to broaden who is represented in AI, and who ultimately benefits from it.
Responsible AI starts with us.
To discover the ultimate winners of the AI4Her awards, join us at the AI4ALL Gala on February 6th 2026, in Amsterdam. The Gala raises vital funds to support our mission to make AI more inclusive, ethical and accessible for everyone. Get your tickets here.
© AI4ALL | 2026