AI Trainer & Evaluator — Aether Generalist (Multimodal)
VideoRLHF
As an AI Trainer at Outlier AI, I evaluated AI-generated responses across text, audio, image, and video. I applied structured rating frameworks to assess accuracy, clarity, safety, and guideline compliance. Factual errors, hallucinations, and misalignments between user instructions and model outputs were identified and documented to improve multimodal model quality.
• Evaluated over 100 AI-generated responses weekly using clear benchmarks.
• Delivered written feedback to inform model training pipelines.
• Conducted prompt analysis and intent interpretation for guideline alignment.
• Accelerated model improvements by surfacing failures and quality gaps.
As an AI Trainer at Outlier AI, I evaluated AI-generated responses across text, audio, image, and video. I applied structured rating frameworks to assess accuracy, clarity, safety, and guideline compliance. Factual errors, hallucinations, and misalignments between user instructions and model outputs were identified and documented to improve multimodal model quality. • Evaluated over 100 AI-generated responses weekly using clear benchmarks. • Delivered written feedback to inform model training pipelines. • Conducted prompt analysis and intent interpretation for guideline alignment. • Accelerated model improvements by surfacing failures and quality gaps.
2025 - 2026