Amardeep Yadav

• Evaluated and validated Large Language Model (LLM) responses for factual accuracy, reasoning quality, safety, and policy compliance.
• Performed prompt engineering by creating, refining, and testing prompts to improve AI model performance across diverse use cases.
• Reviewed AI-generated content for coherence, relevance, instruction adherence, and linguistic quality.
• Annotated and labeled multimodal datasets including text, images, and conversational data to support supervised model training.
• Conducted preference ranking and comparative evaluation of AI outputs using predefined quality guidelines.
• Identified hallucinations, logical inconsistencies, bias, and harmful content, providing structured feedback for model improvement.
• Executed quality assurance (QA) reviews to ensure annotation accuracy and compliance with project specifications.
• Worked with RLHF (Reinforcement Learning from Human Feedback) workflows to improve generative AI model alignment.
• Collaborated with cross-functional teams to resolve edge cases, maintain annotation consistency, and improve evaluation standards.
• Maintained high productivity and accuracy while adhering to strict quality metrics and project SLAs.
• Documented findings and followed detailed annotation guidelines for multiple AI training and evaluation projects.

• Oversaw content safety for short video products, ensuring legal compliance and quality standards.
• Conducted daily audits of short videos to maintain high quality, moderation, and adherence to guidelines.
• Developed and implemented Community Content Security protocols to enhance platform safety.
• Collaborated with Points of Contact (POCs) and the Policy team to identify trends and improve safety policies.
• Created and managed Excel sheets to track daily operations and monitor team performance.
• Analysed reviewed videos for alignment and consistency within the team's objectives.
Organised calibration calls to discuss complex cases, and align best practices across teams.
• Trained and onboarded new team members, facilitating a smooth transition into processes.
• Managed MIS dashboards, KPIs, and daily and weekly performance reports.
Maintained structured documentation for content rights and compliance workflows.
• Coordinated with Policy and Legal teams for content approvals and compliance checklists.