Data Annotation Assistant
I worked on a team training an AI to understand how Nigerians talk. My job was to listen to short voice notes and type out exactly what was said, including pidgin and local slang that i heard. What i did daily: 1. Listened to about 100 clips per day and wrote them out word for word. 2. Labeled each one as question, complaint, greeting, or random talk 3. Marked clips that had bad network, background noise, or mixed Yoruba/Igbo or English 4. Used the client's rulebook to decide edge cases when someone wasn't clear 5. Sent a small report at the end of each day in Excel showing how many i did and any issues I used different tools including Label box for tagging, Audacity to slow down fast talkers, WhatsApp group for quick questions to the lead.