Evaluating Multilingual LLMs at Scale
For Microsoft Research, Karya completed one of the largest multilingual human evaluations of LLMs within three weeks.
Expanding Language Access Through Translation
Karya collaborated with Google to translate over 1.6 million words across 16 long-tail languages, addressing significant challenges such as resource mobilization and technological limitations.
1.6M+
total words
2
novel Unicode Scripts created
16
Long Tail Languages
20+
Districts visited by the team
By engaging local communities and conducting on-the-ground recruitment, Karya built a network of native speakers. The team developed Unicode scripts for languages like Limbu and Lepcha, ensuring digital compatibility. Through targeted training sessions and a focus on cultural nuances, Karya delivered high-quality, contextually relevant translations, showcasing its unique expertise in managing complex linguistic projects in underserved regions.
Related
Conversational Data and Call Center Data Collection(2400 hours)
Collecting and validating conversational speech data across multiple domains in 11 languages for Microsoft