Millicent Ochieng
Research Scientist, Microsoft Research Africa
PhD Researcher in Computer Science, University of the Witwatersrand, Johannesburg
I lead and advise on AI research, strategy, and responsible technology innovation, with a focus on building more equitable AI systems for the Global Majority.
What I work on
Research & Advisory Areas
Applied AI Strategy
Advising organizations on responsible AI adoption, use-case selection, evaluation, governance, and deployment considerations.
Globally Equitable AI
Advancing AI systems that work more fairly and effectively across languages, cultures, and regions, particularly for the Global Majority.
AI Efficiency
Improving model efficiency through multilingual tokenization, better linguistic representation, and reduced downstream compute and cost.
Multicultural AI Alignment
Studying how AI systems represent cultural variation across diverse communities and how alignment can better reflect those differences.
AI Safety
Examining model risks across languages and cultures, including misalignment, bias, reliability, and evaluation failures.
Datasets & Benchmarks
Building culturally grounded datasets and benchmarks for multilingual, low-resource, and underrepresented African settings.
Selected work
Publications
-
EMNLP2026
-
AfricaNLP2026
-
NAACL2025
IrokoBench: A New Benchmark for African Languages in the Age of Large Language Models
★ Outstanding Paper Award, NAACL 2025 ACL Anthology -
AfricaNLP2025
-
NAACL2024
-
EMNLP2023
Recognition
Awards & Fellowships
- 2026Schlumberger Foundation Faculty for the Future FellowshipCompetitive fellowship for women from low- and middle-income countries in STEM
- 2026WiDS High Impact Ambassador, WiDS WorldwideSustained leadership advancing women in data science and AI
- 2025Outstanding Paper Award for IrokoBench, NAACL 2025Outstanding contribution in multilingual and African language NLP
- 2019IUCEA Masters ScholarshipAwarded as the top candidate in Kenya
Beyond the lab
Community & Service
WiDS Kenya
Leading a 400+ member community advancing gender diversity in data science, and serving on the WiDS Worldwide Ambassador Advisory Council.
AfricaNLP
Workshop Organizer & Communications Chair, co-organizing AfricaNLP at EACL 2026.
Masakhane
Advancing multilingual NLP research for low-resource African languages.
ACL
Reviewing submissions in multilingual NLP, large language models and related areas since 2023.
Work with me
Does your AI work effectively, responsibly, and equitably across the people and contexts it is meant to serve?
I help organizations and teams design, build, train, evaluate, and deploy AI systems, while advising on AI strategy, responsible AI, and equitable technology across diverse real-world contexts.