Fine-Tuning Models for Accuracy and Latency at Robinhood Markets - IND392

Explore a comprehensive conference talk from AWS re:Invent 2025 that demonstrates how Robinhood Markets successfully enhanced their AI performance through strategic fine-tuning of large language models using Amazon SageMaker and Amazon Bedrock. Learn how this financial technology company achieved remarkable results, including 80% lower latency on critical agentic tasks while scaling their operations to process millions of tokens per minute. Discover practical techniques for fine-tuning foundation models with Amazon SageMaker, implementing high-performing inference solutions on Amazon Bedrock, and applying effective model selection strategies. Gain insights into optimization approaches and deployment best practices that can help your team improve AI application performance on AWS infrastructure. The session provides real-world case study examples from Robinhood's implementation, covering the technical challenges they faced and the solutions they developed to overcome latency and accuracy issues in their AI-powered systems.