AI Model Evaluation Analyst at SparkAI
Image
Bounding BoxEvaluated model accuracy for live visual data streams from autonomous hardware to ensure correct obstacle identification and safe behavior in real-world operations. Trained and assessed multiple AI models under different logic and safety rules, adapting evaluation to each model’s project requirements. Maintained high quality targets and provided audit-focused insights to improve overall model performance and reliability. • Measured and validated model performance against accuracy and safety standards for autonomous obstacle detection. • Trained six separate AI models with distinct behavior rules, ensuring no cross-contamination of safety logic. • Sustained a 99% accuracy rate over approximately three and a half years while supporting ongoing reliability improvements. • Contributed insights that supported implementation of an internal audit system to increase team accuracy.