How Lyft’s ML platform speeds ML development
Despite a 286% increase in total runs (execution of a data/ML task), and 424% increase in production runs, Lyft’s AWS-based ML platform LyftLearn cost 6% less in 2022 than in 2021
Unstoppable surge of generative AI
The generative AI cosmos is expanding at warp speed, with its foundation models, middleware, and applications coalescing to create an interconnected, dynamic landscape. In this thriving ecosystem, middleware emerges as a critical piece, seamlessly linking foundational models with applications by providing bespoke fine-tuning, optimization, and integration services. Pioneers like Hugging Face lead the charge, offering a rich array of pre-trained models, tools, and libraries.A dazzling array of end-user solutions across various use-cases, from content generation and chatbots to image synthesis, are coming on stream. A Marketsandmarkets.com projection puts the global AI market at a staggering $191 billion by 2025.Among other deep dives this week in NPW Insights, get a preview of the generative AI landscape.
Insights
CNCF Whitepaper: Platforms for cloud-native
Platforms are crucial in cloud-native paradigms because they are able to separate supporting, non-differentiating capabilities like DBs, message queues, brokers, observability collectors, and authentication systems from app-specific logic more than in previous paradigms. The whitepaper looks at defining attributes of platforms, like composability and self-service, and then organizes components of a platform into two categories: platform interfaces, like portals, APIs, CLIs, and golden path templates, and platform capabilities like development environments, observability, infrastructure, data, messaging, and event services, and secrets management capabilities. Each of these components are also mapped to relevant CNCF and CDF projects. Also useful: three impact areas to measure success of platforms, and specific KPIs for each area; how to organize and enable platform teams. WHITEPAPER
