Brian Kimaiga

Key Skills

Software

No software listed

Top Subject Matter

Conversational AI

LLM dialogue testing

LLM prompt engineering and response evaluation

Top Data Types

Text

Video

Image

Top Task Types

Text Generation

Freelancer Overview

Lead Dialogue Writer and Evaluator (Multi-Persona AI Dialogue Simulation Project). Brings 8+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Mindrift, Internal, and Proprietary Tooling. Education includes Bachelor of Arts, Kenyatta University (2021). AI-training focus includes data types such as Text and labeling workflows including Evaluation and Rating.

Labeling Experience

Lead Dialogue Writer and Evaluator (Multi-Persona AI Dialogue Simulation Project)

MindriftText

Led multi-persona LLM dialogue simulation to stress-test AI conversational performance. Designed structured datasets labeling hallucination, inconsistency, emotional misreading, and contradiction in model responses. Developed evolving test prompts to probe performance under constraint escalation. • Stress-tested LLMs against multi-turn, persona-rich conversational scripts • Created structured labels for hallucination and reasoning error analysis • Analyzed dialogue logs and annotated model response quality • Iteratively improved annotation and evaluation rubrics

Led multi-persona LLM dialogue simulation to stress-test AI conversational performance. Designed structured datasets labeling hallucination, inconsistency, emotional misreading, and contradiction in model responses. Developed evolving test prompts to probe performance under constraint escalation. • Stress-tested LLMs against multi-turn, persona-rich conversational scripts • Created structured labels for hallucination and reasoning error analysis • Analyzed dialogue logs and annotated model response quality • Iteratively improved annotation and evaluation rubrics

2024 - Present

Prompt Engineer (LLM Prompt Engineering and Response Evaluation Project)

MindriftText

Developed and applied prompt-based evaluation protocols for multiple LLMs. Built structured rubrics for clarity, consistency, task completion, and safety scoring of LLM responses. Documented model differences and best practices for high-accuracy prompting. • Created and tested over 200 prompts spanning creativity, memory, and reasoning • Evaluated response quality across leading commercial models • Authored detailed assessments for prompt-response coherence and reliability • Identified and documented optimal prompt strategies for targeted model improvement

Developed and applied prompt-based evaluation protocols for multiple LLMs. Built structured rubrics for clarity, consistency, task completion, and safety scoring of LLM responses. Documented model differences and best practices for high-accuracy prompting. • Created and tested over 200 prompts spanning creativity, memory, and reasoning • Evaluated response quality across leading commercial models • Authored detailed assessments for prompt-response coherence and reliability • Identified and documented optimal prompt strategies for targeted model improvement

2023 - 2024

AI Dialogue Tester and Content Writer

Text

Simulated persona-driven dialogues to evaluate LLM consistency and performance. Conducted adversarial and emotional response testing to annotate failure patterns. Authored structured reports detailing error types and recommendations. • Created diverse dialogue scripts to stress-test AI behavior • Conducted granular labeling of model outputs (consistency, instruction-following, emotion) • Identified and documented subtle LLM failure modes • Modified prompts and reporting for evolving test cases

Simulated persona-driven dialogues to evaluate LLM consistency and performance. Conducted adversarial and emotional response testing to annotate failure patterns. Authored structured reports detailing error types and recommendations. • Created diverse dialogue scripts to stress-test AI behavior • Conducted granular labeling of model outputs (consistency, instruction-following, emotion) • Identified and documented subtle LLM failure modes • Modified prompts and reporting for evolving test cases

2022 - 2024

Education

Work History