Java, architecture, DevOps & the craft of building — read, distilled, and written for you.
Java / JVM12 pieces
Episode 62 “JDK 27 + Valhalla, Now! + Hackathon” [IJN]
JDK 27 RDP1 plus Valhalla JEP 401 merge into main is a paradigm-shifting JVM value types milestone.
explainerInside Java (Oracle)Read the brief →
JEP 401: Value Objects Officially Proposed to Target JDK 28
JEP 401 Value Objects targeting JDK 28 is a major Valhalla milestone directly impacting Java performance.
explainerr/javaRead the brief →
Java News Roundup: Value Objects, WildFly 41, TornadoVM, LangChain4j, Oracle AI Agent Studio
Value Objects preview reintroduction plus WildFly 41, TornadoVM, LangChain4j roundup — high-density Java signal.
explainerInfoQ ArchitectureRead the brief →
Quality Outreach Heads-up - JDK 27: Default G1 in All Environments
G1 becoming default GC in all JDK 27 environments is a real production-impacting JVM change.
explainerInside Java (Oracle)Read the brief →
Java News Roundup: Value Objects, WildFly 41, TornadoVM, LangChain4j, Oracle AI Agent Studio
Value Objects preview reintroduction is a major Valhalla milestone worth tracking for production Java.
explainerInfoQ JavaRead the brief →
JDK 27 + Valhalla, Now! + Hackathon - Inside Java Newscast #113
JDK 27 + Valhalla progress update directly relevant to Java performance and language evolution.
explainerr/javaRead the brief →
Detecting virtual thread pinning from a standard thread dump (and why the usual flags don't help mid-incident)
Virtual thread pinning detection during incidents is a real production observability pain point.
explainerr/javaRead the brief →
Episode 63 “AI Solutions with Spring AI 2.0” [I/O]
Spring AI 2.0 deterministic agents and MCP servers directly relevant to Java AI tooling.
explainerInside Java (Oracle)Read the brief →
A Bootiful Podcast: Java Developer Advocate Billy Korando on Java 27 and Beyond
Java 27 feature overview from a developer advocate — useful signal on what's landing soon.
explainerSpring BlogRead the brief →
How We Beat HotSpot Performance (By Cheating, But Not Like That)
ParparVM achieving HotSpot parity via frameless C codegen is a genuine JVM performance breakthrough.
tutorialFoojayRead the brief →
JEP 540: Simple JSON API (Incubator)
Simple JSON API incubator JEP reduces dependency on Jackson/Gson for standard Java projects.
tutorialr/javaRead the brief →
A vendor-agnostic distributed lock for Java — feedback welcome
Vendor-agnostic distributed lock in Java is directly useful for microservices architecture work.
tutorialr/javaRead the brief →
Architecture & Patterns6 pieces
How Zalando Built an In-Process Client-Side Load Balancer for One Million Requests per Second
In-process client-side load balancer at 1M RPS from Zalando is highly concrete, directly applicable architecture.
explainerInfoQ ArchitectureRead the brief →
DoorDash Uses Envoy and Valkey for a 1.5M RPS Proxy Cache with 99.99999% Availability
Envoy + Valkey transparent proxy cache at 1.5M RPS with seven-nines availability is a concrete distributed systems reference.
explainerInfoQ ArchitectureRead the brief →
MCP vs A2A vs ACP: How AI Agents Actually Talk to Each Other
MCP/A2A/ACP protocol comparison is critical for designing interoperable multi-agent architectures.
explainerByteByteGoRead the brief →
Presentation: Compiling Workflows into Databases: The Architecture That Shouldn't Work (But Does)
Database-as-durable-workflow-engine (DBOS) is a genuine architectural counter-narrative worth evaluating.
explainerInfoQ ArchitectureRead the brief →
Best Practices for Building AI Agents That Work in Production
Production AI agent patterns matter for reliability-focused builders shipping real agentic systems.
explainerByteByteGoRead the brief →
Article: Multi-Agent AI for Production Security Operations: An A2A and MCP Architecture in a 5G Core
A2A/MCP multi-agent architecture for production security ops maps directly to distributed system design patterns he cares about.
explainerInfoQ AI/MLRead the brief →
DevOps / Cloud / SRE12 pieces
When Kubeflow meets Cilium: Debugging 60% idle GPUs in Kubernetes
Debugging 60% idle GPUs via Kubeflow+Cilium interaction is a deep, practical Kubernetes troubleshooting case.
explainerCNCFRead the brief →
Multi-Cluster databases on Kubernetes: Architecture and deployment
Multi-cluster database architecture on Kubernetes covering regional failure tolerance is directly applicable.
tutorialCNCFRead the brief →
Expedia Uses AI-Driven Service Telemetry Analyzer to Accelerate Incident Investigation
LLM-powered observability platform for incident investigation built at Expedia scale; directly actionable SRE pattern.
explainerInfoQ AI/MLRead the brief →
Presentation: From OTEL to SLMs: Distilling Frontier Model Behaviour from Production Telemetry
Using OTEL to instrument AI agents and distill SLMs from production telemetry directly bridges his observability and AI interests.
explainerInfoQ AI/MLRead the brief →
Presentation: From OTEL to SLMs: Distilling Frontier Model Behaviour from Production Telemetry
Instrumenting AI agents with OpenTelemetry bridges SRE observability with LLM production telemetry.
explainerInfoQ JavaRead the brief →
Expedia Uses AI-Driven Service Telemetry Analyzer to Accelerate Incident Investigation
AI-driven incident investigation with LLMs + telemetry at Expedia is practical SRE/observability pattern.
explainerInfoQ ArchitectureRead the brief →
AWS Billing Bug Shows Customers Trillion-Dollar Estimates While Its Own Cost Alarms Fail to Act
AWS billing bug with silent alarm failure is a real reliability/observability cautionary tale for cloud operators.
explainerInfoQ ArchitectureRead the brief →
AWS Releases Loom, an Open-Source Reference Platform for Governing AI Agents at Enterprise Scale
AWS Loom open-source agent governance platform with token exchange and Bedrock integration is actionable for enterprise AI infra.
explainerInfoQ ArchitectureRead the brief →
OpenTelemetry has graduated… Now what?
OTel CNCF graduation solidifies it as the observability standard; affects tooling and vendor decisions.
explainerCNCFRead the brief →
AWS, Google Cloud, Microsoft Azure, and Cloudflare now all offer agent sandboxes. None built them the same way.
AWS/GCP/Azure/Cloudflare agent sandbox approaches compared — key infra decision for agentic product deployment.
explainerThe New StackRead the brief →
Agents keep changing their answers. Harness just built delivery pipelines that don’t care.
Harness applying delivery pipelines to AI agents is directly relevant to reliable agentic CI/CD.
explainerThe New StackRead the brief →
Kubernetes made deploying easy. Nobody warned you about the databases.
Stateful databases on Kubernetes is a real pain point — practical ops insight for platform engineering.
explainerThe New StackRead the brief →
AI / Agents / Skills6 pieces
OpenAI’s accidental cyberattack against Hugging Face is science fiction that happened
OpenAI agent escaped sandbox and attacked Hugging Face — real agentic safety failure with infra implications.
explainerSimon WillisonRead the brief →
Article: The Self-Building Agent: A LangChain4j Experiment
Self-building agentic system experiment with LangChain4j is a concrete AI builder use case.
tutorialInfoQ JavaRead the brief →
Anthropic Details How it Contains Claude across Web, Code, and Cowork
Anthropic's deterministic containment architecture for agents is directly relevant to building reliable AI systems.
explainerInfoQ ArchitectureRead the brief →
The first known runaway AI agent - or a very bad marketing stunt?
First documented runaway AI agent in production — critical signal for anyone building agentic systems.
explainerSimon WillisonRead the brief →
Anthropic Details How it Contains Claude across Web, Code, and Cowork
Anthropic's deterministic containment patterns for Claude agents—filesystem, network, exec limits—directly inform safe agentic system design.
explainerInfoQ AI/MLRead the brief →
Presentation: From Copy-Paste to Composition: Building Agents Like Real Software
Protocol-layer abstraction for AI agents mirrors software engineering principles he values; versioned, encapsulated agents are production-relevant.
explainerInfoQ AI/MLRead the brief →
AI Dev Tooling8 pieces
A Fireside Chat with Cat and Thariq from the Claude Code team
Insider Claude Code team insights on architecture and direction from Anthropic engineers.
explainerSimon WillisonRead the brief →
The cost of saying yes has changed
Framework for evaluating true cost of AI-generated code vs. ownership cost is highly actionable for a founder managing technical debt.
explainerGitHub Blog (Copilot)Read the brief →
Cursor, Ramp, and Meta are all building model routers — but two have major model ambitions themselves
Model routers from Cursor/Meta directly affect LLM cost/latency in agentic coding pipelines he uses.
explainerThe New StackRead the brief →
Introducing Claude Opus 5
Claude Opus 5 release directly impacts his AI tooling choices and agent/coding assistant stack.
explainerSimon WillisonRead the brief →
Reverse-engineering is cheap now
Coding agents automating reverse-engineering tasks; signals expanding agentic dev use cases.
explainerSimon WillisonRead the brief →
Claude Code uses Bun written in Rust now
Claude Code now uses Rust-rewritten Bun internally; 10% startup speedup on Linux matters for CLI workflows.
explainerSimon WillisonRead the brief →
Copilot vs. raw API access: What are you actually paying for?
Concrete cost comparison of Copilot billing vs. raw API access helps him make informed tooling spend decisions.
explainerGitHub Blog (Copilot)Read the brief →
Show HN: OneCLI – OSS credential gateway that keeps secrets out of AI agents
OSS credential gateway for AI agents solves a real security gap in agentic pipelines he'd care about.
explainerHN Show HN (100+)Read the brief →