Favour Edache

Key Skills

Software

Scale AI

Appen

Surge AI

Top Subject Matter

General LLMs

Medical and Technical Content

Reasoning Tasks

Top Data Types

Text

Document

Top Task Types

Prompt + Response Writing (SFT)

Freelancer Overview

AI Training Specialist | Prompt Verifier. Core strengths include Scale AI, Appen, and Surge AI. Education includes Bachelor of Science (2024). AI-training focus includes data types such as Text and labeling workflows including Evaluation, Rating, and Prompt + Response Writing (SFT).

Labeling Experience

Hallucination Detection Initiative

Text

In the Hallucination Detection Initiative, I designed a framework to systematically identify unsupported claims in AI-generated content. I created an error taxonomy adopted by annotation teams for prioritization. This project enhanced the identification and classification of hallucinations in medical and technical model outputs. • Developed error-type taxonomy for annotation and verification • Applied framework to ongoing evaluation of AI-generated outputs • Focused on medical and technical subject areas for accuracy • Improved annotation precision for complex scenarios

In the Hallucination Detection Initiative, I designed a framework to systematically identify unsupported claims in AI-generated content. I created an error taxonomy adopted by annotation teams for prioritization. This project enhanced the identification and classification of hallucinations in medical and technical model outputs. • Developed error-type taxonomy for annotation and verification • Applied framework to ongoing evaluation of AI-generated outputs • Focused on medical and technical subject areas for accuracy • Improved annotation precision for complex scenarios

2024 - Present

Prompt Optimization Series Portfolio Project

TextPrompt Response Writing SFT

During the Prompt Optimization Series project, I redesigned ambiguous prompts to enhance response precision in reasoning-intensive tasks. I documented changes with before-and-after comparisons, quantifying measurable improvements. My work improved instruction clarity and overall prompt effectiveness in LLM training. • Revised and clarified 50+ complex prompts • Demonstrated measurable gains in model performance via documentation • Contributed to clarity and precision within LLM training datasets • Supported prompt engineering best practices for annotation teams

During the Prompt Optimization Series project, I redesigned ambiguous prompts to enhance response precision in reasoning-intensive tasks. I documented changes with before-and-after comparisons, quantifying measurable improvements. My work improved instruction clarity and overall prompt effectiveness in LLM training. • Revised and clarified 50+ complex prompts • Demonstrated measurable gains in model performance via documentation • Contributed to clarity and precision within LLM training datasets • Supported prompt engineering best practices for annotation teams

2023 - Present

AI Training Specialist | Prompt Verifier

Scale AIText

As an AI Training Specialist and Prompt Verifier, I evaluated and ranked large language model responses for accuracy, clarity, and safety using standardized metrics. I detected hallucinations and ambiguities in model outputs, providing actionable corrective feedback. I contributed to reducing model error rates through dataset improvements and prompt template refinement. • Evaluated and rated 500+ model responses weekly using gold-standard guidelines • Flagged and documented edge cases and factual inaccuracies for model retraining • Partnered with engineering teams on prompt optimization to decrease ambiguous outputs • Supported a 15% reduction in model error rates across a 6-month period

As an AI Training Specialist and Prompt Verifier, I evaluated and ranked large language model responses for accuracy, clarity, and safety using standardized metrics. I detected hallucinations and ambiguities in model outputs, providing actionable corrective feedback. I contributed to reducing model error rates through dataset improvements and prompt template refinement. • Evaluated and rated 500+ model responses weekly using gold-standard guidelines • Flagged and documented edge cases and factual inaccuracies for model retraining • Partnered with engineering teams on prompt optimization to decrease ambiguous outputs • Supported a 15% reduction in model error rates across a 6-month period

2023 - Present

Education

Work History