Required Skills & Experience
• Senior-level experience delivering end-to-end ML systems (problem definition → data strategy → model development → evaluation → deployment)
• Strong computer vision expertise, specifically object detection models (YOLO required)
• Hands-on experience using LLMs via API (GPT-4o or similar) in production or near-production environments
• Experience with multimodal AI (image + text analysis)
• Proven ability to work with multilingual text data
• Experience building solutions in compliance-sensitive environments (accuracy, documentation, traceability)
• Strong understanding of model evaluation metrics (false positives/negatives, edge case handling)
Nice to Have Skills & Experience
• Lead end-to-end development of AI models for sustainability claim verification
• Build and train YOLO-based logo detection models, including labeling strategy and data quality iteration
• Develop and implement LLM-based verification workflows using GPT-4o API for text and multimodal analysis
• Define and refine claim verification logic and model evaluation methodology
• Work with datasets including artwork/images, sustainability claims, multilingual text, and compliance data
• Collaborate with internal stakeholders to define what qualifies as a "verifiable claim" and required evidence
• Deliver a production-ready pipeline and support integration into business workflows (e.g., dashboards or tools)
• Document methodology, testing results, and deployment recommendations
Job Description
• Senior-level experience delivering end-to-end ML systems (problem definition → data strategy → model development → evaluation → deployment)
• Strong computer vision expertise, specifically object detection models (YOLO required)
• Hands-on experience using LLMs via API (GPT-4o or similar) in production or near-production environments
• Experience with multimodal AI (image + text analysis)
• Proven ability to work with multilingual text data
• Experience building solutions in compliance-sensitive environments (accuracy, documentation, traceability)
• Strong understanding of model evaluation metrics (false positives/negatives, edge case handling)