AI/LLM Analyst (Contract) – Innodata (Client: Meta)
As part of Meta’s multimodal AI evaluation, I conducted video annotation involving segmentation tasks to support actions and object labeling within video frames. My focus was on quality assurance and instruction compliance in video-labeling assignments. I contributed to optimizing text-to-video AI evaluation and hallucination identification. • Labeled video data for segmentation and action recognition. • Ensured high accuracy in frame-by-frame annotation. • Evaluated VLM alignment in multi-modal video tasks. • Detected inconsistencies and maintained labeling standards throughout.