Vision+LLM Multi-modal Labeling Specialist
ImageClassification
Developed image classification pipelines using EfficientNet-B3 for custom datasets of 2,000 labeled images deployed for document understanding and visual question answering. Managed manual and automated labeling workflows to achieve target accuracy and reduction in review time. Used Gradio for interactive validation of model predictions and label consistency.
• Supervised manual and semi-automated image labeling tasks
• Coordinated workflow across vision and document QA domains
• Ensured dataset quality for transfer learning applications
• Achieved review time savings of 20 hours per week
Developed image classification pipelines using EfficientNet-B3 for custom datasets of 2,000 labeled images deployed for document understanding and visual question answering. Managed manual and automated labeling workflows to achieve target accuracy and reduction in review time. Used Gradio for interactive validation of model predictions and label consistency. • Supervised manual and semi-automated image labeling tasks • Coordinated workflow across vision and document QA domains • Ensured dataset quality for transfer learning applications • Achieved review time savings of 20 hours per week
2024 - Present