Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Hi, I’m Raphael Muselema, an AI Data Trainer and Prompt Engineer with 19 months of hands-on experience in evaluating large language models. I specialize in creative alignment, visual context auditing, and multi-turn preference ranking, bringing an engineering mindset to digital design and multimodal evaluation. I excel at translating complex model outputs into structured, aesthetically coherent results through UI/UX considerations, Python/JavaScript tooling, and rigorous grounding checks. I am fully available for remote work and collaborate effectively with cross-functional teams to deliver reliable, user-friendly AI systems.…Hi, I’m Raphael Muselema, an AI Data Trainer and Prompt Engineer with 19 months of hands-on experience in evaluating large language models. I specialize in creative alignment, visual context auditing, and multi-turn preference ranking, bringing an engineering mindset to digital design and multimodal evaluation. I excel at translating complex model outputs into structured, aesthetically coherent results through UI/UX considerations, Python/JavaScript tooling, and rigorous grounding checks. I am fully available for remote work and collaborate effectively with cross-functional teams to deliver reliable, user-friendly AI systems.

Raphael Muselema

AI Prompt Engineer, AI Developer, Data Scientist, +2





Hi, I’m Raphael Muselema, an AI Data Trainer and Prompt Engineer with 19 months of hands-on experience in evaluating large language models. I specialize in creative alignment, visual context auditing, and multi-turn preference ranking, bringing an engineering mindset to digital design and multimodal evaluation. I excel at translating complex model outputs into structured, aesthetically coherent results through UI/UX considerations, Python/JavaScript tooling, and rigorous grounding checks. I am fully available for remote work and collaborate effectively with cross-functional teams to deliver reliable, user-friendly AI systems.…Hi, I’m Raphael Muselema, an AI Data Trainer and Prompt Engineer with 19 months of hands-on experience in evaluating large language models. I specialize in creative alignment, visual context auditing, and multi-turn preference ranking, bringing an engineering mindset to digital design and multimodal evaluation. I excel at translating complex model outputs into structured, aesthetically coherent results through UI/UX considerations, Python/JavaScript tooling, and rigorous grounding checks. I am fully available for remote work and collaborate effectively with cross-functional teams to deliver reliable, user-friendly AI systems.

Available to hire

Hi, I’m Raphael Muselema, an AI Data Trainer and Prompt Engineer with 19 months of hands-on experience in evaluating large language models. I specialize in creative alignment, visual context auditing, and multi-turn preference ranking, bringing an engineering mindset to digital design and multimodal evaluation.
I excel at translating complex model outputs into structured, aesthetically coherent results through UI/UX considerations, Python/JavaScript tooling, and rigorous grounding checks. I am fully available for remote work and collaborate effectively with cross-functional teams to deliver reliable, user-friendly AI systems.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Work Experience

Independent AI Evaluator (LLM SxS) at jContract via Micro1 & Mercor

February 1, 2025 - Present

Execute blind SxS preference ranking for complex, multi-turn conversational tasks, specifically evaluating models on tone, creative alignment, and formatting. Audit simulated retrieval panes and visual context windows to ensure natural integration of information without over-personalization or forced inferences. Draft structured, 75–150 word, turn-referenced rationales to justify model rankings using specific textual and visual evidence snippets. Maintain a 92% per-task QA acceptance rate against a strict 85% project passing threshold, measured via weekly random QA audits. Achieve a rolling 8-week calibration score with an 89% preference ranking agreement against adjudicated gold labels.

Technical Prompt Engineer & UI Developer at jPrimeGrade Solutions

January 1, 2025 - Present

Design structured master prompts utilizing few-shot methodologies to force LLMs into strict formatting and creative constraints. Iterate prompt designs based on documented failure modes (e.g., formatting breaks, context loss) to establish reliable generation workflows for highly structured outputs. Develop custom UI/UX elements and front-end layouts using HTML and CSS for web platforms, successfully bridging technical AI outputs with engaging visual design.

Data Annotation Specialist at Independent Contractor

July 1, 2024 - January 1, 2025

Annotated and classified 5,000+ text and data points for early-stage conversational AI models, supporting rigorous machine learning training pipelines. Verified generated outputs against provided source documents for factual grounding, tone consistency, and negative constraint adherence. Categorized defect frequencies related to logical flow and missing constraints, maintaining a QA reversal rate of less than 5% against a 10% allowable project baseline.