At Character.AI, I review model-generated and user-generated content for safety violations, including edge cases involving minors, exploitation, and policy evasion. I also label and escalate high-risk interactions to support moderation model refinement and minor safety standards.
I work with cross-functional teams to test products, features, and updates, identify safety-control failures and policy gaps, and improve product and moderation outcomes. I evaluate moderation quality across internal teams, BPO partners, and system behavior, then turn findings into clearer policy guidance.
At Meta, I processed reports involving takedowns, appeals, retractions, and suspicious reporting. I updated workflow policies and processes, with policy simplification efforts updating or removing 40% of policy and workflow information and increasing efficiency and turnaround times.
I also led internal and vendor training and quality initiatives at Meta, and authored quality guidelines and consistency checks. This work maintained a 93% quality score, raised the consistency score from 93% to 100%, improved misuse detection accuracy from 13% to 36%, and improved recall to 67%.

