Steven Otieno

Trained and evaluated advanced language models on complex theological, historical, and comparative religion scenarios to identify failure modes and bias, focusing on accuracy, neutrality, and cultural sensitivity. Produced reproducible error traces, structured metacognitive feedback, and high-quality annotated datasets spanning Christianity, Islam, Hinduism, Buddhism, Judaism, and African Traditio

Rated and ranked AI-generated creative and factual content for quality, accuracy, and coherence across literature, history, culture, and humanities. Wrote adversarial prompts and provided detailed preference annotations to improve RLHF training data, delivering grammatical, stylistic, and factual corrections while participating in calibration sessions.

Annotated NLP training corpora for sentiment analysis, named entity recognition (NER), intent classification, and coreference resolution in English text. Ensured annotation consistency via inter-annotator agreement checks and calibration, and delivered schema-compliant JSON outputs while flagging culturally sensitive or ambiguous constructions for review.

Evaluated web search results for relevance and user intent alignment, assessing misinformation, harmful language, and adherence to quality guidelines. Reviewed speech-to-text and transcription accuracy for voice recognition datasets, maintained 20+ tasks/hour, and achieved quality scores above 95%, including feedback for localization and East African English variants.

Provided academic writing support across English literature, religious studies, linguistics, social work, and healthcare management, helping undergraduate and postgraduate students with research paper structure, source evaluation, and APA/MLA/Chicago citation formatting. Delivered original journal entries, case studies, and reflective essays aligned to instructor rubrics and built a 30+ client bas

Moderated online community forums and live chat channels by enforcing guidelines and acting on reports for hate speech, harassment, misinformation, and policy violations. Maintained response times under 3 minutes during peak traffic, achieved a 98% community satisfaction score, and escalated complex cases with evidence documentation while updating playbooks with trust & safety teams.

Scored K–8 students’ oral reading fluency by analyzing recordings against standardized literacy rubrics. Identified phonemic, phonological, and comprehension errors to inform intervention plans, provided feedback to teachers within 24 hours, and maintained accuracy above 97% across thousands of assessments.

Facilitated weekly adult literacy sessions for 15–20 participants, focusing on functional English reading, writing, and communication skills. Developed customized learning materials tailored to learners’ varied literacy levels and occupational needs.

Grade: Second Upper Division

Activities and societies: Active member of the KU English Literary Society and the Debate and Oratory Club.

Earned a Bachelor of Arts in English and Linguistics, graduating in February 2020. Specialized in discourse analysis, sociolinguistics, applied linguistics, phonetics & phonology, literary theory, and translation studies, including a final-year research project on code-switching in urban Kenyan youth communication.