Projects
- ai-engineering-from-scratchBuild AI engineering from the ground up. A full course in code.
- agentmemoryPersistent memory for AI coding agents, powered by iii-engine.
- pro-workflowSelf-correcting Claude Code workflows. Hooks, agent teams, the 80/20 ratio.
- awesome-claude-code-toolkitThe most comprehensive toolkit for Claude Code: agents, skills, commands, plugins.
- devops-interview-questionsA central hub of DevOps interview questions to help you land your next job.
- skillkitUniversal CLI for managing AI agent skills across 40+ coding agents.
- k8sgamesLearn Kubernetes by playing. Real kubectl in a 3D browser game, no install.
- DevOpsCommunityA community to learn and explore DevOps with experienced professionals.
- kubectl-mcp-serverMCP server for Kubernetes with 270+ tools. In the CNCF Landscape.
- CreateMVPCreate MVPs with AI in seconds.
- awesome-openclawA curated list for the OpenClaw agentic AI ecosystem.
- manim-video-generatorGenerate Manim math animations from natural language prompts.
- File Share ManagerShare and manage files on any cloud storage with a clean UI.
Writing
- Handwired: How JavaScript Becomes an Exact Neural Net
- TypeSafe Raised $870M. Can Jev See? 408 Calls on 48 Real Photos
- agentgateway in Production: One Endpoint for MCP Servers, Agents, and Models
- Continue Is Not Consent: How a Harness Nudge Became Permission
- Build a Personal RAG Harness: Jev Judges, a Local Model Writes
- Laya and Jev: System 1 Models Answer With a Type, Not Text
- Agent Substrate Egress and Its Gaps: Nothing the Agent Says Is Trusted
- Agent Substrate: How Suspended Agents Share a Warm Pool
- Google AX (Agent Executor): Task, Workspace, Gateway, Model
- Inside the Devin Harness: Two Loops and a Handoff
- SWE-Bench Pro V2 Caught Two Models Cheating. The Fix Was Moving the Grader.
- Opus 5.5 in Practice: Effort Is the Only Dial Left
- Amazon Blocked Muse. The Request Had No Name on It.
- Zed Turned Off Pull Requests. The Merge Gate Has to Go Somewhere.
- BITCOS: Intel's Ternary LLM Weight Format Goes Below 1.58 Bits
- MCP Servers Can Now Ship Their Own Manual
- The Harness Is an API Now. Here Is What You Still Own.
- The Eval Sandbox Held the Keys
- RSA-260 Fell to an Agent Fleet. The Human Sent 3,328 Messages.
- GPT-6 Astra Changed the Shape of a Turn
- Four Flash Models in Four Months. The Bill Per Task Went Up 40%.
- Fable 5.1 Burned the Meter in Twenty Minutes
- Fable 5.1 and Mythos 5.1: One Model, Two Doors
- Claude Has Two Meters. The 20x Is Only on One.
- OpenAI Cut Cursor Off. Five Percent Is the Interesting Number.
- The Model Hardware Standard: Read, Write, and the Plate That Must Be There
- The Bulletin Board: How Isolated Agents Built a Message Bus Out of a Package Registry
- MCP's New Roadmap: Events Without Sessions, Identity Without API Keys
- Grok Bot: The Agent Got a Computer. Your Bots Share It.
- DFlash 2: The Drafter Stopped Guessing Alone
- All About the A2A Protocol, Now Part of the Agentic AI Foundation
- Claude Has Levels. Most People Stop at Two.
- The AGENTS.md Practices Nobody Uses
- The Harness, Not the Model
- The Inference Engine Underneath
- Fifty Startups Coming Next
- Stateless MCP: The Protocol Deleted Its Own Handshake
- OpenWiki: The LLM Wiki Pattern as Running Code
- The LLM Wiki: Stop Retrieving, Start Compounding
- LLM Wiki v2: What Breaks at Scale
- The Inference Bill: Why AI's Best Customers Lose the Most Money
- The Forward Deployed Engineer: The Job the AI Industry Paid Billions to Rediscover
- What a Forward Deployed Engineer Actually Does
- Graph Engineering Is 290 Years Old
- Inside the Pi Harness: The Agent That Shows You the Meter
- Inside the Claude Code Harness: The System Prompt Machine
- Inside the Codex Harness: Compaction as an API Primitive
- Kimi CLI and K3: The Newest Harness Has the Strangest Ideas
- Tokens per Second Is a Memory Bandwidth Number
- The AGENTS.md Genre
- The Four Loops, Clearly Explained
- The Composable Harness: How Agents Actually Get Reliable
- New Names, Old Fundamentals: Decoding the Agent Era's Rebrands
- Your Codebase Has a Second Reader Now
- Your Row Changed Hook Isn't CDC: The Dual-Write Trap
- Harness Engineering in Practice: Stop Agents Building the Wrong Thing
- Harness Engineering: The Real Leverage Is Around the Model
- How to Become an AI Engineer in 2026: A Complete Roadmap
- Loops, Graphs, and the Layer That Matters
- Loop Engineering Is Just Software Engineering
- The 2026 AI Engineering Roadmap
- How to Build Your Own Agent Harness
- The Substrate Is Installable
- Why Agent Sandboxes Are Converging on libkrun, Not Firecracker
- Building Agents for the Real World
- Your AI Agent's Memory Is Probably Broken. Here Is Why.
- I Ran Six Agent Sandbox Runtimes Back to Back
- The Harness Is the Backend
- Three Primitives Are Enough: Rebuilding Backends Without Fifty SDKs
- WebMCP: Chrome Turned Every Website Into an API for AI Agents
- OpenClaw: 180K Stars, a Critical CVE, and Agentic AI's Wake-Up Call
- Claude Code: The Hooks, Skills, and MCP Workflow That Ships Software
- OpenCode: The 95K-Star AI Coding Agent Built for the Terminal
- Moltbook: The Social Network Where 1.5M AI Agents Post and Scheme
- AI Coding Agents Compared: Claude Code vs OpenCode vs OpenClaw
- MCP vs A2A vs Agent Skills: The Complete AI Agent Protocol Stack
- MCP vs A2A: The Complete Guide to AI Agent Protocols
- The Ralph Loop: The Most Important AI Coding Technique of 2026
- One Primitive for Your Entire Backend
- Why DevRel Is Needed for Your Company
- My Take on Cursor AI After Using It 5× a Day
- DevOps Roadmap 2025
- A Complete Guide to Next.js Authorization
- How to Implement Authorization in React.js
- Implementing Cerbos in a Spring Boot Application
- Creating a Service Directory with Backstage.io
- A Closer Look at Botkube ChatOps for Kubernetes
- Top 5 Resiliency Trends of 2023
- Exploring Kubernetes 1.28 Sidecar Containers
- Using Kubernetes with AWS Lambda: Scaling Serverless Apps
- Istio Service Mesh Cheatsheet
- Zero Trust for Cloud Native Applications
- Kubernetes from Scratch (series)
- Learn Terraform for Free
Guides
- Evals and BenchmarksRead a benchmark score, then build, grade, calibrate, and hillclimb your own eval for any model or agent harness.
- Inside the TransformerAttention built from a masked average, nine years of diffs to the 2026 block, then sampling and serving. Thirteen interactive figures.
- Inside MilesOne RL step through RadixArk's Miles: the refit session, weight transports, async staleness, and the sampler-trainer gap, with four interactive figures and measured Mac runs.
- Inside SGLangOne prompt across CPU and GPU, real attention matrices, the radix tree, HiCache tiers, and the scheduler, with eight interactive figures, measured Mac runs, and GPU runs.
- Inside the vLLM EngineThe block pool, the phase-free scheduler, speculative decoding, and grammar masking, with six interactive figures.
- What Actually Fits on Your MachineInteractive fit calculator: 55 models, quant and KV math, offload speed estimates.
- Tokens per Second Is a Memory Bandwidth NumberThe speed side of the same arithmetic: watch a decode step, fold a model, predict your tok/s.
- Run Muse Glimmer LocallyMeta's day-one 30B agent model: envelope packer, exact GGUF sizes, DFlash drafter, honest benchmarks.
- Run DeepSeek V4 Flash LocallyQuant picker rewrites every command on the page for your machine.
- All local model guidesKimi K3, Qwen3, gpt-oss, Gemma 3, Mistral, Llama 4, GLM, and more.
Talks
- MCP, Open Standards & Data in the Age of Agentic AI
- Securing AI Workloads: Zero-Trust Architecture for LLM Applications
- Building Event-Driven Workflows for Cloud Native Applications
- Building Event-Driven Workflows for Cloud Native Applications
- Kubernetes Community Days UK (speaker)
- Orchestrating LLM Inference with Kubernetes
- Chaos & Behind the Scenes of Kubernetes in Production
- Enhanced Platform Engineering: Infrastructure Modernization
- Embracing Multi-Tenancy While Scaling MLOps
- Securing Service Meshes with eBPF
- Kubernetes Observability with eBPF
Committees
- Exam Contributor: Certified Cloud Native Platform Engineering Associate
- Program Committee: KubeCon + CloudNativeCon China
- Program Committee: KubeCon + CloudNativeCon Japan
- Program Committee: State of OpenCon
- Exam Contributor: Certified Backstage Associate
- Program Committee: Cilium & eBPF Day
- Program Committee: MultitenancyCon
- Program Committee: Kubernetes Community Days
- Program Committee: eBPF Summit