AWS Machine Learning Blog
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications
Bayer Launches Multi-Agent Engine on AgentCore: 80% of Questions Handled Without Engineers
Bayer Crop Science built ADE on Amazon Bedrock AgentCore: 22 agents handle about 80% of questions without an engineer, and failure diagnosis dropped from an hour to 10 minutes.
AWS Launches Multi-Turn RL in SageMaker: gpt-oss-20b Climbs From 15% to 69% for $339
SageMaker AI now fine-tunes agents on a single reward for an entire chain of steps. On Wordle, gpt-oss-20b rose from 61% to 84% of words solved; training tokens cost $339.
Property Finder Found a Flaw in Its Labels and Cut VLM Costs by 86%
96.8% of Property Finder's reference labels were inherited from its old model. After relabeling, Llama 4 Maverick on Bedrock scored Micro F1 0.80 vs 0.56 at $2.35 per 1000 images.
AWS adds agent payments to AgentCore: x402 support in days instead of months
Amazon Bedrock AgentCore payments lets AI agents pay for services per request, with spending limits enforced by the infrastructure. x402 support now takes days, not months.
AWS's Path-to-Value Framework: Getting AI Into Production
Amazon Web Services has introduced Path-to-Value (P2V), a structured approach meant to help businesses turn AI concepts into real products that generate steady revenue.
Amazon Launches Optimized AI Model Deployment in SageMaker JumpStart
A new Amazon SageMaker JumpStart feature offers preconfigured machine learning deployment setups tuned for specific use cases and performance requirements.
Amazon SageMaker HyperPod: Optimizing AI Model Inference and Cutting Costs
Amazon SageMaker HyperPod offers tools for running generative AI inference. The platform cuts costs by up to 40% and speeds up neural network deployment in the enterprise.
OpenAI GPT-6 Sol, Luna and Astra Now Available in Bedrock From Five Asian Regions
AWS has opened three OpenAI GPT-6 models via global cross-Region inference in Bedrock for Singapore, Jakarta, Thailand, Malaysia and Taipei.
AWS shows how to run a single LLM across two GPU nodes on EKS with Ray LLM DLC
AWS Machine Learning Blog details deploying one LLM on two Amazon EKS GPU nodes: prefill and decode run on separate nodes, with a single OpenAI-compatible endpoint exposed to clients.
AWS shows how to run llama.cpp on SageMaker and get probabilities over a set of options
AWS published a guide to deploying llama.cpp on SageMaker AI: an OpenAI-format chat API plus SemIf, an extension that returns probabilities over predefined options on CPU or GPU.
Postman brings Agent Mode to 40 million developers on Amazon Bedrock
AWS and Postman detail Agent Mode's architecture: three techniques to curb tool sprawl and context bloat on Amazon Bedrock for 40 million users.
AWS September Roundup: Bedrock, AgentCore and Strands Get Updates
AWS rounded up its September updates to Amazon Bedrock, AgentCore and Strands: more models, serverless agents with built-in evaluation, and auto-syncing knowledge bases.
AWS Publishes Blueprint for Sharing a Single HyperPod EKS Cluster Across Teams
AWS has described a reference architecture for one SageMaker HyperPod cluster on EKS: five layers, from IAM Identity Center sign-in to per-namespace cost tracking.
AWS Professional Services: How to reduce project timelines from months to days
AWS ProServe rebuilt its delivery process from within and compressed engagement timelines from months to days — not by layering AI on top of legacy processes, but by reconstructing everything from scratch.
Amazon Nova 2 Sonic: How to Build a Voice Agent for Patient Scheduling
AWS has published a guide to building a voice agent on Nova 2 Sonic that reminds patients of visits, confirms appointments, and reduces no-show rates.
AWS: mathematical optimization solves what intuition can't
AWS Innovation Center explains how mathematical optimization helps businesses make precise decisions with thousands of variables — and why neither intuition nor conventional ML can cope here.
Yahoo Plugs Amazon Bedrock Into Search Retargeting on Its Yahoo DSP Ad Platform
Yahoo has integrated Amazon Bedrock into Yahoo DSP's search retargeting system. AI now finds users with the right search intent across Yahoo's own properties and partner sites.
monday.com Puts AI Agents Into Production on Amazon Bedrock: 50%+ More PRs per Engineer
monday.com runs AI agents in production on Amazon Bedrock: 9 in 10 developers use AI for code every month, and PR output per engineer is up more than 50%.
Amazon SageMaker AI Learns to Stream Model Benchmark Results to MLflow
Amazon added streaming MLflow integration to SageMaker AI: benchmark job metrics, parameters, and charts now flow into a unified experiment tracker in real-time.
AWS Demonstrated Five Resilience Patterns for Generative AI on Amazon Bedrock
AWS Machine Learning Blog Described Five Practical Resilience Patterns for GenAI Applications on Amazon Bedrock — from Native Service Functions to LLM Gateway
AWS demonstrated infrastructure for multi-turn RL training of Amazon Nova model
AWS published a guide for deploying event-driven infrastructure for multi-turn RL training of the Amazon Nova model on Amazon's SageMaker HyperPod platform.
Claude Sonnet 5 from Anthropic became available on Amazon Bedrock and AWS
Anthropic released Claude Sonnet 5—the first Sonnet model of the new generation—and opened access to it through Amazon Bedrock and Claude Platform on AWS.
AWS Demonstrates Generative UI for AI Agents on Bedrock AgentCore via AG-UI
AWS explained how the AG-UI protocol and CopilotKit create interactive interfaces for AI agents on Amazon Bedrock AgentCore platform.
AWS Simplified Access to Amazon Bedrock Models via Managed Entitlements
AWS introduced managed entitlements for Bedrock—a single subscription to models from a central account with access distribution across the organization.
Outpost VFX Accelerated AI Model Training for Visual Effects 8x on AWS
Outpost VFX studio achieved 8x speedup in AI model training for facial replacement on AWS infrastructure.
IBS Software Built Bilingual AI NER Model for Cargo Logistics on Bedrock
IBS Software demonstrated how it built a bilingual named entity recognition model for cargo logistics on Amazon Bedrock.
AWS Releases Open-Source Amazon Bedrock Model Profiler for Model Selection
AWS introduced open-source Amazon Bedrock Model Profiler—a tool aggregating model metadata from multiple sources into a single search interface.
AWS Demonstrates BoltzGen Protein Design Model Deployment on SageMaker AI
AWS Machine Learning Blog showed how to deploy the BoltzGen model for protein design on Amazon SageMaker AI—from quick validation to production workloads.
AWS: Fine-Tuning Amazon Nova Models Increased Email Data Extraction Accuracy to 94.77%
AWS demonstrated a case study: fine-tuning Amazon Nova models through Amazon SageMaker AI raised email data extraction accuracy to 94.77% and reduced processing costs by 50%.
AWS Explains How It Safely Launches Frontier AI Models in Bedrock Service
Amazon Web Services has shared its approach to safely launching frontier AI models for customers through a managed cloud service Amazon Bedrock.