Ian Gonzalez

I am an LLM training specialist with hands‑on experience designing high‑volume prompts, building evaluation rubrics, and improving model pass@1 and incident rates. I combine technical annotation tooling knowledge with practical automation to speed analysis and increase consistency.

At Welocalize I produced 2,000+ prompts per month, ran 300+ adherence tests, and helped raise pass@1 by 4.8 percentage points while reducing incidents by 19%. I have also led A/B comparisons, authored evaluation rubrics, mentored evaluators, and created reusable prompt templates as a freelance prompt engineer.

My work emphasizes safety and measurable improvement — from proposing safety lock‑policies and curating a 150‑item Safety Prompt Bank to automating sampling and weekly reporting to cut analysis time. I am fluent in Spanish and advanced in English, and I use tools like Google Sheets, basic Python, regex, and Jira to support evaluation and reporting.