Abhiram User

I’m an AI Safety & Trust Analyst with hands-on experience evaluating large-scale LLM outputs for safety, quality, and policy compliance. I’ve maintained 96% accuracy and achieved a 99% audit pass rate across high-volume annotation workflows.

I assess AI-generated outputs for hallucinations, bias, and unsafe responses, while strengthening human-in-the-loop feedback loops through prompt-response analysis and annotation quality control. I also trained 25+ associates and improved annotation accuracy from ~70% to 85%+ by enforcing consistent evaluation standards.

I focus on data integrity and reliability—conducting root cause analysis, reducing data rework by 15%, and escalating high-complexity edge cases to protect dataset quality. I’m proficient in Python and SQL for validation and workflow optimization, and I want to contribute to improving model reliability, alignment, and safety.