At Meta, I annotate Swahili text, speech transcripts, and multimodal data for language-model and speech-recognition training. I maintain 98% or higher accuracy against internal gold sets and flag edge cases that have led to updates to the official taxonomy.
At Google, I labeled Swahili speech audio and text for speech-to-text and natural language understanding models. I also quality-checked peer annotations and documented informal Swahili and regional dialect cases for guideline refinement.
At Amazon (Alexa AI), I annotated Swahili utterances for intent classification, slot filling, and entity recognition in voice assistant training. My background includes a Bachelor of Arts in Linguistics from University of Nairobi and a Diploma in Information Technology from Kenyatta University.
