InfoQ Homepage Architecture & Design Content on InfoQ
-
AI-Powered Bot Compromises GitHub Actions Workflows across Microsoft, DataDog, and CNCF Projects
AI-powered bot hackerbot-claw exploited GitHub Actions workflows across Microsoft, DataDog, and CNCF projects over 7 days using 5 attack techniques. Bot achieved RCE in 5 of 7 targets, stole GitHub token from awesome-go (140k stars), and fully compromised Aqua Security's Trivy. Campaign included first documented AI-on-AI attack where bot attempted prompt injection against Claude Code.
-
Cloudflare Releases Experimental Next.js Alternative Built with AI Assistance
Cloudflare released vinext, an experimental Next.js reimplementation built on Vite by one engineer, with AI guidance over one week, for $1,100. Early benchmarks show 4.4x faster builds, but Cloudflare cautions it's untested at scale. Missing static pre-rendering. HN reaction skeptical, noting Vite does the heavy lifting. Already running on CIO.gov despite experimental status.
-
Netflix Automates RDS PostgreSQL to Aurora PostgreSQL Migration across 400 Production Clusters
Netflix engineers describe an internal automation platform that migrates nearly 400 RDS PostgreSQL clusters to Aurora, reducing downtime and operational risk. The platform coordinates replication, CDC handling, controlled cutover, and rollback, while supporting service teams in a self-service migration workflow.
-
Java News Roundup: Apache Solr 10, LangChain4j, Grails, JobRunr, Gradle, Devnexus, Commonhaus
This week's Java roundup for March 2nd, 2026, features news highlighting: the GA release of Apache Solr 10; point releases of LangChain4j, JobRunr, Multik and Gradle; maintenance releases of Grails and Keycloak; Devnexus 2026; and Pi4J joining the Commonhaus Foundation.
-
Scaling Human Judgment: How Dropbox Uses LLMs to Improve Labeling for RAG Systems
To improve the relevance of responses produced by Dropbox Dash, Dropbox engineers began using LLMs to augment human labelling, which plays a crucial role in identifying the documents that should be used to generate the responses. Their approach offers useful insights for any system built on retrieval-augmented generation (RAG).
-
Standardizing Post-Quantum IPsec: Cloudflare Adopts Hybrid ML-KEM to Replace Ciphersuite Bloat
Cloudflare has extended hybrid post-quantum encryption to IPsec and WAN traffic, standardizing its SASE stack ahead of the NIST 2030 deadline. By adopting a streamlined ML-KEM key exchange, the move addresses long-standing "ciphersuite bloat" in quantum-resistant IPsec. The update aims to neutralize "harvest now, decrypt later" threats without requiring specialized hardware upgrades.
-
Architecting for Global Scale: inside DoorDash’s Unified, Composable Dasher Onboarding Platform
DoorDash has rebuilt its Dasher onboarding into a unified, modular platform to support global expansion. The new architecture uses reusable step modules, a centralized status map, and workflow orchestration to ensure consistent, localized onboarding experiences. This design reduces complexity, supports market-specific variations, and enables faster rollout to new countries.
-
OpenAI Secures AWS Distribution for Frontier Platform in $110B Multi-Cloud Deal
OpenAI's $110B funding includes AWS as the exclusive third-party distributor for the Frontier agent platform, introducing an architectural split: Azure retains stateless API exclusivity; AWS gains stateful runtime environments via Bedrock. Deal expands the existing $38B AWS agreement by $100B and commits 2GW of Trainium capacity.
-
Decentralizing Architectural Decisions with the Architecture Advice Process
Our system architectures have changed as technology and development practices have evolved, but the way we practice architecture hasn’t kept up. According to Andrew Harmel-Law, architecture needs to be decentralized, similar to how we have decentralized our systems. The alternative to having an architect take and communicate decisions is to “let anyone make the decisions” using the advice process.
-
From Central Control to Team Autonomy: Rethinking Infrastructure Delivery
Adidas engineers describe shifting from a centralized Infrastructure-as-Code model to a decentralized one. Five teams autonomously deployed over 81 new infrastructure stacks in two months, using layered IaC modules, automated pipelines, and shared frameworks. The redesign illustrates how to scale infrastructure delivery while maintaining governance at scale.
-
AWS Launches Agent Plugins to Automate Cloud Deployment
AWS launched Agent Plugins for AWS, providing AI coding agents with specialized deployment skills. The initial deploy-on-aws plugin transforms workflows by accepting commands like "deploy to AWS" and generating complete pipelines with architecture recommendations, cost estimates, and infrastructure code. Supported in Claude Code and Cursor, AWS claims 10-minute deployments versus hours manually.
-
Google Enhances Node Pool Auto-Creation Speed for GKE Clusters
Google Cloud has optimised GKE's node pool auto-creation, significantly cutting "Time to Ready" for massive clusters. By improving control plane communication and request batching, GKE now provisions resources faster, rivalling tools like Karpenter. The update enhances scaling reliability and stability for high-volume AI and batch workloads, automatically rolling out across supported versions.
-
Hybrid Cloud Data at Uber: How Engineers Solved Extreme-Scale Replication Challenges
Uber’s HiveSync team optimized Hadoop Distcp to handle multi-petabyte replication across hybrid cloud and on-premise data lakes. Enhancements include task parallelization, Uber jobs for small transfers, and improved observability, enabling 5x replication capacity and seamless on-premise-to-cloud migration.
-
Java News Roundup: Lazy Constants, TornadoVM 3.0, NetBeans 29, Quarkus, JReleaser, Open Liberty
This week's Java roundup for February 23rd, 2026, features news highlighting: new JEP 531 Candidate, Lazy Constants; GA releases of TornadoVM 3.0 and NetBeans 29; point releases of Quarkus, JReleaser, Chicory and RefactorFirst; maintenance releases of Micronaut and Jox; and the February 2026 edition of Open Liberty.
-
Pinterest’s CDC-Powered Ingestion Slashes Database Latency from 24 Hours to 15 Minutes
Pinterest launched a next-generation CDC-based database ingestion framework using Kafka, Flink, Spark, and Iceberg. The system reduces data availability latency from 24+ hours to 15 minutes, processes only changed records, supports incremental updates and deletions, and scales to petabyte-level data across thousands of pipelines, optimizing cost and efficiency.