At Micro1, I evaluate, compare, and rank AI-generated responses for accuracy, reasoning, relevance, and instruction adherence. I identify hallucinations and other errors, then provide written feedback to support model improvement.
I also assess multimodal AI, including robotics performance, object manipulation, and instruction following. I use timestamps and frame-by-frame video and audio review to verify performance and activity completion.
Previously, at Outlier AI, I evaluated AI-generated content across multiple workflows, applying project guidelines and scoring systems. I also studied LLM evaluation, human feedback, guardrails, and open-source models through DeepLearning.AI professional development.
