An AWS re:Invent chalk talk on responsible AI in the era of large language models, with a focus on the practical challenges teams face when bringing generative AI into production. The session covered key responsible AI considerations such as fairness, privacy and security, explainability, robustness, governance, and transparency, along with common risks including hallucinations, toxicity, data leakage, and intellectual property concerns. I also discuss practical mitigation strategies and scalable architecture patterns, including grounded LLMs and self reflective RAG, for building safer and more reliable AI applications.

Slides