Omachoko Y.

Key Skills

Software

Mercor

Micro1

Mindrift

Other

Top Subject Matter

AI/LLM evaluation and data annotation quality assurance

Cybersecurity, AI safety evaluation and labeling

Reinforced Learning with Human Feedback

Top Data Types

Text

Document

Audio

Top Task Types

Red Teaming

RLHF

Fine-tuning

Computer Programming/Coding

Freelancer Overview

Brings hands-on professional experience in AI safety evaluation, technical diagram assessment, data annotation, and LLM trajectory validation. Core strengths include RL Studio evaluations, adversarial safety audits, multi-stage AI output validation, and structured technical analysis. Education includes Bachelor of Engineering, Ahmadu Bello University (2008 – 2015) and Introduction to Digital Currencies, Blockchain Technology and Cryptocurrencies (Non-degree), University of Nicosia (Grade: 88.33%). AI-training focus includes data types such as Images, Computer Code, Programming tasks, and technical documentation; with labeling workflows including Evaluation, Quality Rating (1–5 scale), Safety Assessment, and RLHF.

Labeling Experience

AfterQuery Expert (Data Annotator)

OtherTextRLHF

Validated AI trajectory outputs through multi-stage rubric-based checks for determinism, alignment, and technical accuracy. Identified and mitigated ambiguity signals by tightening constraints and refining gold-standard answers. Curated and standardized professional benchmark artifacts and authored SWE-bench-style coding tasks for model capability evaluation. • Graded complex rubric outputs to ensure factual and technical correctness. • Reduced ambiguity by reframing questions and updating high-value reference answers. • Standardized benchmark materials (DOCX, PPTX, XLSX) for consistent evaluation. • Built automated test harnesses to verify generated code against reference implementations.

2026 - Present

Mercor

ImageRLHF

• Assigned quality ratings to technical issues (Major vs Minor) and checked diagrammatic elements. • Conducted adversarial safety audits to flag dangerous content, hate speech, and graphic violence. • Verified technical data and spelling against authoritative references to ensure label fidelity. • Executed structured evaluation workflows with a focus on reliability for downstream AI systems.

2026 - Present

Education

Work History