Open to opportunities

John Waititu

@johnwaititu

Message

Detail-oriented AI data annotator specializing in multilingual transcription and speech dataset quality.

Kenya

Message

What I'm looking for

I’m looking to support AI research teams by building safer, more accurate models through high-quality, guideline-compliant multilingual dataset work—speech, transcription, and content evaluation—while continuously improving annotation efficiency and dataset accuracy.

I’m a detail-oriented AI Data Annotator with more than four years of experience delivering high-quality datasets for NLP, speech, and computer vision projects. I annotate text, audio, images, and video with consistent accuracy rates between 96 and 100 percent, and I focus on multilingual transcription in English and Swahili, guideline compliance, and quality assurance.

I’m especially strong in speech and content workflows: I’ve transcribed and time-stamped over 200 hours of multilingual audio interviews, and I’ve translated and localized over 150,000 words between Swahili and English for NGOs, theses, apps, and YouTube subtitles. From my work reviewing and scoring over 2,000 AI outputs weekly with A/B testing and policy-violation flagging, I help teams refine model behavior and build safer, more accurate systems.

Experience

Work history, roles, and key accomplishments

Current

Freelance Linguist Specialist

Current

Upwork

May 2025 - Present (1 year)

Transcribed and time-stamped 200+ hours of Swahili, Sheng, Kikamba, Luhya, and English interviews, focus groups, and podcasts with 98–100% accuracy. Annotated speech data in Praat/ELAN/FLEX and translated/localized 150,000+ words for NGOs, academic work, apps, and subtitles, with consistent quality and guideline compliance.

Transcription Time Stamping Linguistic Annotation Praat ELAN Flex Translation Localization Content Evaluation Quality Assurance

Freelance Data Annotator

Appen

Dec 2022 - Apr 2025 (2 years 4 months)

Reviewed, scored, and ranked 2,000+ AI-generated outputs per week to improve NLP model performance while maintaining 96%+ accuracy and weekly quotas. Conducted 300+ A/B tests to assess clarity/timing/stylistic alignment and flagged unsafe or policy-violating content, reducing recurrence by 15%.

A B Testing Content Safety Policy Labeling Quality Assurance Workflow Optimization

Research Assistant

Mount Kenya University

Oct 2011 - Dec 2012 (1 year 2 months)

Conducted literature reviews and organized 50–100 references per month using Zotero, including structured summaries for academic work. Performed fieldwork and weekly transcription/translation of 2–4 hours of audio, and supported analysis and academic writing using Praat/ELAN, Excel, and R.

literature review Zotero Fieldwork Transcription Praat ELAN