Organizations and individuals running multiple custom AI models, especially recent Mixture of Experts (MoE) model families, can face the challenge of paying for idle GPU capacity when the individual models don’t receive enough traffic to saturate a dedicated compute endpoint. To solve this problem, we have partnered with the vLLM community and developed an efficient …
Something has shifted in the developer community over the past year. AI agents have moved from “interesting research concept” to “thing my team is actually building.” The prototypes are working. The demos are impressive. And now comes the harder question: How do we ship this? That question turns out to be a multi-part one. Agents …
Read more “A developer’s guide to production-ready AI agents”
Large Language Models (LLMs) can be adapted to extend their text capabilities to speech inputs. However, these speech-adapted LLMs consistently underperform their text-based counterparts—and even cascaded pipelines—on language understanding tasks. We term this shortfall the text-speech understanding gap: the performance drop observed when a speech-adapted LLM processes spoken inputs relative to when the original text-based …
Read more “Closing the Gap Between Text and Speech Understanding in LLMs”
Managing large photo collections presents significant challenges for organizations and individuals. Traditional approaches rely on manual tagging, basic metadata, and folder-based organization, which can become impractical when dealing with thousands of images containing multiple people and complex relationships. Intelligent photo search systems address these challenges by combining computer vision, graph databases, and natural language processing …
Reasoning and planning are the bedrock of intelligent AI systems, enabling them to plan, interact, adapt, and ultimately, operate independently. At Apple, understanding and advancing reasoning capablilities in AI systems has long been an area of active research, and has resulted in numerous publications that both explore new techniques to advance the frontier of reasoning, …
Avneesh Saluja, Santiago Castro, Bowei Yan, Ashish Rastogi Introduction Netflix’s core mission is to connect millions of members around the world with stories they’ll love. This requires not just an incredible catalog, but also a deep, machine-level understanding of every piece of content in that catalog, from the biggest blockbusters to the most niche documentaries. As …
Read more “MediaFM: The Multimodal AI Foundation for Media Understanding at Netflix”
Critical labor shortages are constraining growth across manufacturing, logistics, construction, and agriculture. The problem is particularly acute in construction: nearly 500,000 positions remain unfilled in the United States, with 40% of the current workforce approaching retirement within the decade. These workforce limitations result in delayed projects, escalating costs, and deferred development plans. To address these …
Read more “Scaling data annotation using vision-language models to power physical AI systems”
In 2025, Amazon SageMaker AI saw dramatic improvements to core infrastructure offerings along four dimensions: capacity, price performance, observability, and usability. In this series of posts, we discuss these various improvements and their benefits. In Part 1, we discuss capacity improvements with the launch of Flexible Training Plans. We also describe improvements to price performance …
As artificial intelligence and machine learning (AI/ML) workflows grow in scale and complexity, it becomes harder for practitioners to organize and deploy their models. AI projects often struggle to move from pilot to production. AI projects often fail not because models are bad, but because infrastructure and processes are fragmented and brittle, and the original …
Read more “Build AI workflows on Amazon EKS with Union.ai and Flyte”
Nearly every snowboard trick carries a number. A 1080 means three full rotations. A 1440 means four. The convention is simple: add up every rotation around every axis and count in 180° increments. For decades it’s served as the sport’s universal shorthand for difficulty. Judges, coaches, and athletes all speak this language fluently. It’s also, …
Read more “Using Google Cloud AI to measure the physics of U.S. freestyle snowboarding and skiing”