AI Knowledge Assistant Creator & Evaluator, Allstate Insurance
Text
Built and evaluated a governed AI assistant (Leo) for decision support inside Allstate. Led accuracy checks and error correction on AI model outputs in claims and policy logic. Developed validation controls and human-in-the-loop workflows to ensure AI decisions matched expert understanding.
• Identified and documented model hallucinations in insurance outputs.
• Translated insurance workflows into structured training data for AI.
• Piloted evaluation with 20 claims operations staff for iterative improvement.
• Established rigorous feedback loops for claims and policy AI behavior.
Built and evaluated a governed AI assistant (Leo) for decision support inside Allstate. Led accuracy checks and error correction on AI model outputs in claims and policy logic. Developed validation controls and human-in-the-loop workflows to ensure AI decisions matched expert understanding. • Identified and documented model hallucinations in insurance outputs. • Translated insurance workflows into structured training data for AI. • Piloted evaluation with 20 claims operations staff for iterative improvement. • Established rigorous feedback loops for claims and policy AI behavior.
2025 - Present