Trainer/Educator: Multimodal AI Data Collection & Annotation

The Role

The Trainer will lead training on multimodal AI datasets involving synchronized image, video, audio, text and sensor data. The role includes instruction on multimodal annotation, metadata management, dataset preparation and LLM/GenAI annotation workflows.

Responsibilities:

• Deliver training on multimodal AI data collection, annotation methodologies, annotation standards, and industry best practices for developing high-quality AI datasets.
• Train participants in synchronized multimodal data capture, including the collection and management of image, video, audio, text, and sensor data, while emphasizing synchronization techniques and data quality requirements.
• Conduct practical sessions on image, video, audio, text, and cross-modal annotation, covering object detection, segmentation, object tracking, speech and speaker annotation, named entity recognition (NER), sentiment analysis, and vision-language alignment using industry-standard annotation tools.
• Teach metadata management, dataset organization, annotation quality assurance (QA), validation techniques, and standard dataset formats such as COCO, YOLO, JSON, and XML to prepare learners for real-world AI data engineering workflows.
• Introduce learners to Generative AI and LLM data annotation workflows, including response ranking, preference labeling (RLHF-style), hallucination and factuality detection, safety annotation, and synthetic data quality validation.
• Mentor and supervise learners in executing industry-based multimodal AI capstone projects, guiding them through data collection, annotation, quality review, documentation, and project presentation while ensuring adherence to industry standards and best practices.

Qualification:

• Master's/ Ph.D in AI, Computer Science, Robotics or Data Science.
• 3+ years' experience in AI data operations.
• Experience with multimodal AI datasets.