Leigh G.

Designed, tested, and refined advanced prompts to train generative models under SFT, RLHF, RLFAIF and related frameworks; maintained prompt libraries, best practices, and internal guidelines to ensure consistent output quality. Promoted to Quality Analyst to evaluate model outputs for accuracy, safety, cultural nuance, and bias; recommended and implemented improvements based on qualitative and quantitative performance data. Partnered with language specialists and R&D teams to run experiments, analyze results, and iteratively improve model behavior. Contributed to internal docs and training materials and joined client meetings to align deliverables with stakeholder objectives. Clients included: OpenAI, Cohere, AI21 Labs, xAI, Amazon, and Inflection AI.

Supported multi-client generative AI projects as part of large global training and evaluation programs. Designed and refined prompts to improve tone, clarity, factuality, and safety in LLM outputs; documented patterns and proposed improvements in UX and linguistics. Advised startups on prompt strategy, content rules, and AI-generated content quality; aligned prompts with brand voice and technical constraints in collaboration with linguists, UX and engineering teams. Contributed to trust & safety evaluations, including red-team analyses and model-behavior analyses, for clients such as OpenAI, Cohere, AI21 Labs, xAI, Amazon, and Inflection AI.

Led UX writing, product copy, brand language, and content design for consumer-facing websites and mobile applications. Improved product flows and crafted interface copy, product names, navigational labels, taxonomies, and user education materials across web and app experiences. Delivered pitches and content design projects for government, fintech, healthcare, e-commerce, and food sector clients (Enterprise Singapore, Mandai Wildlife Group, NTU, Pharmaniaga, UniFi, AEON Bank).