At Amazon, I audited structured and semi-structured data using SQL and JSON, and evaluated multimodal outputs for Amazon Nova Sonic and Nova Reel against project guidelines.
I led a 10-member content refinement team that achieved 98% SLA compliance through QA protocols, dialogue evaluations, and guideline adherence. I also conducted root cause analyses on labeling errors and partnered with product managers and AI engineering teams on quality issues.
I assembled a 12-person pilot team for a JSON-based workflow and supported trainers in preparing SOPs; the customer later made the project permanent. I'm also developing a Python project to audit LLM annotation quality, building on my data quality experience at Amazon.
