Toluwani O.

Key Skills

Software

Other

Top Subject Matter

LLM evaluation and creative writing generation quality assurance

RLHF evaluation and video action annotation for computer vision

Top Data Types

Text

Video

Image

Top Task Types

Action Recognition

Classification

Text Generation

Text Summarization

Fine-tuning

Transcription

Evaluation/Rating

Question Answering

Freelancer Overview

AI Prompt Specialist (Human-in-the-Loop evaluator and prompt chain refinement) at Rovic Oil and Gas Limited (Remote). Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other. Education includes Bachelor of Science, University of Lagos (UNILAG) (2025). AI-training focus includes data types such as Text and Video and labeling workflows including Evaluation, Rating, and Action Recognition.

Labeling Experience

Freelance AI Data Specialist & Evaluator (Mindrift RLHF, AtlasCapture video annotation, Micro1 prompt specialist)

OtherVideoAction Recognition

Conducted reinforcement learning from human feedback (RLHF) tasks including editing model responses for logic, factuality, and style adherence as part of human-in-the-loop refinement. Completed video data annotation by identifying temporal actions and ensuring precise timestamping for computer vision training. Delivered prompt stress-testing support using complex chain-of-thought prompt strategies to evaluate LLM reasoning robustness. • Edited and validated agent responses for logic, factuality, and style compliance. • Performed RLHF human review to refine agent behaviors. • Annotated video segments with action timing and accurate timestamps. • Designed complex prompt chains to stress-test LLM reasoning capabilities.

2025 - Present

AI Prompt Specialist (Human-in-the-Loop evaluator and prompt chain refinement) at Rovic Oil and Gas Limited (Remote)

OtherText

Performed human-in-the-loop evaluation of LLM outputs by reviewing responses for weaknesses, unclear decision points, and edge cases. Conducted devil’s advocate red-teaming by stress-testing models with contradictory prompts to verify consistent policy adherence and realism. Produced and refined prompt chains to improve output quality, coverage, and reduce revision cycles. • Reviewed AI-generated creative writing for logic, intent adherence, and policy compliance. • Identified and documented 20+ edge cases and created gold-standard examples. • Iterated solutions based on evaluation findings to improve precision and coverage. • Coordinated requirement clarification with stakeholders to maintain alignment.

2025 - 2025

Education

Work History