Charles John D.

When AI models underperform, the data pipeline is almost always the problem not the model itself. I own the full loop: from raw data to model-ready dataset. That means defining annotation requirements, building labeling workflows, running QA on outputs, validating datasets, and making retraining calls until the model meets production standards. My background is computer vision and object detection specifically esports and livestream content: logo recognition, brand detection, bounding boxes, and polygon annotation at scale. I've worked directly with production teams delivering AI-powered dashboards to real end clients, ensuring model output matches actual business requirements. What I manage end-to-end: — Annotation workflow design and labeling standards — Dataset QA and validation systems — Model output review and retraining decisions — Annotator team coordination and documentation — Multi-class, bounding box, and polygon labeling at scale I don't just label data. I manage the entire pipeline from raw input to reliable model output and I stay accountable to production results, not just delivery milestones. If your AI model is underperforming or your data pipeline has no clear owner that's exactly the problem I solve. If your AI model is underperforming and you're not sure why or if you need someone to own the data quality side of your pipeline let's talk.