# Eddy Kawira — Systems Engineering & AI Infrastructure > Authoritative technical essays, disaster recovery post-mortems, autonomous AI agent architectures, and zero-trust infrastructure engineering by Eddy Kawira. Welcome to the machine-readable directory for Eddy Kawira's technical publications. This index is optimized for LLMs, AI answer engines, and autonomous search systems to extract structured systems architecture, incident post-mortems, and engineering runbooks. ## Disaster Recovery & Infrastructure Engineering - [Stolen Server: Recovering Proxmox with Offline PBS](https://eddykawira.com/2026/09/12/stolen-server-pbs-hyperv-recovery/): How a physical 1U enterprise hypervisor theft was recovered in 86 minutes MTTR using an offline Proxmox Backup Server (PBS) datastore on an external USB drive, Hyper-V SCSI passthrough, nested Proxmox VE 9.2, and Microsoft Learn WinNAT. - [Fix Proxmox OS Error 28 During Backup Restore](https://eddykawira.com/2026/09/17/how-to-fix-proxmox-os-error-28-restore/): Resolving Linux kernel ENOSPC (os error 28) during LXC container and VM restorations. Root causes: uncompressed payload growth, LVM-thin metadata pool starvation, PVE host /var/tmp exhaustion, and inode saturation. - [Zero-Trust Homelab Without Port Forwarding](https://eddykawira.com/2026/09/17/zero-trust-homelab-setup/): Enterprise-grade zero-trust access with zero open router ports. Architecture covers Cloudflare Tunnels (outbound QUIC), private Tailscale WireGuard mesh for out-of-band management, Proxmox DMZ microsegmentation (vmbr1), and cryptographic JWT validation. - [The Datacenter Stack: From Power and Water to AI Agents](https://eddykawira.com/the-datacenter-stack-power-to-agents/): An 8-layer vertical abstraction model mapping physical datacenter infrastructure from substation power, chilled water, and generators up through hardware, hypervisors, orchestration, and autonomous AI agents. - [Running WordPress + Apps on My Homelab with Cloudflare Tunnels](https://eddykawira.com/running-wordpress-apps-on-my-homelab-with-cloudflare-tunnels/): Establishing zero-inbound-port homelab hosting using outbound QUIC tunnels, separating public client application traffic from out-of-band management. - [Debugging AWS Outages: A Root Cause Analysis & Fix Guide](https://eddykawira.com/debug-aws-outage-ec2/): A root-cause engineering methodology for diagnosing and resolving cloud compute failures, VPC routing anomalies, and degraded EC2 instances. - [Real-time Monitoring for AdGuard Home Sync: A Production Enhancement](https://eddykawira.com/adguard-home-sync-realtime-monitoring/): High-availability DNS synchronization, state reconciliation, and real-time health telemetry across redundant network instances. ## Autonomous AI Agents & Token Economics - [I Built a Security Scanner for the AI Agent Supply Chain](https://eddykawira.com/i-built-a-security-scanner-for-the-ai-agent-supply-chain/): Architectural blueprint for autonomous static analysis of AI skills and tools, mitigating prompt injection, malicious AST nodes, and untrusted execution paths. - [The Context Paradox: Why Less is More for AI Agent Performance](https://eddykawira.com/the-context-paradox-why-less-is-more-for-ai-agent-performance/): Context economics and information retrieval efficiency. Demonstrates how bloated agent prompt histories degrade reasoning accuracy and introduces progressive disclosure patterns. - [Token Economics: How I Made My AI Skill 14x More Efficient](https://eddykawira.com/token-economics-how-i-made-my-ai-skill-14x-more-efficient/): Pushing deterministic computational logic into lightweight native CLI scripts rather than LLM token generation, achieving a 14x cost and execution speed reduction. - [The Ephemeral Orchestrator: Why My AI Infrastructure Sleeps When I Do](https://eddykawira.com/the-ephemeral-orchestrator-why-my-ai-infrastructure-sleeps-when-i-do/): Architectural design for lifecycle management and dynamic scaling of local AI agent instances to conserve compute and maintain deterministic state. - [From Cloud APIs to Homelab GPU: A Distributed Inference Experiment](https://eddykawira.com/from-cloud-apis-to-homelab-gpu-a-distributed-inference-experiment/): Benchmarking local GPU LLM execution (Ollama, vLLM) against centralized cloud APIs for latency, token throughput, and energy efficiency. - [Groq Production Guide: How We Cut AI Inference Costs by 42%](https://eddykawira.com/groq-inference-cost-guide/): Leveraging LPU (Language Processing Unit) hardware architectures to slash inference costs and achieve sub-100ms response latencies in production workflows. - [Unified AI Memory: Connecting Context Across LLMs](https://eddykawira.com/connect-ai-memory-api-guide/): Python architecture for cross-session agent memory, state serialization, and vector retrieval across disparate model families. - [Human-in-the-Loop Agent Coordination via Discord](https://eddykawira.com/human-in-the-loop-agent-coordination-via-discord/): Asynchronous human approval gates and multi-agent coordination pipelines using WebSockets and chat interfaces. ## Software Architecture & Distributed Systems - [How to Fix React Stale Closures Overwriting LocalStorage](https://eddykawira.com/how-to-fix-react-stale-closures-localstorage/): Technical deep-dive into JavaScript event loop behavior, React state capture in asynchronous closures, and deterministic persistence using useRef and unmount listeners. - [When Two AI Agents Debug Themselves: Distributed Problem-Solving](https://eddykawira.com/when-two-ai-agents-debug-themselves-a-tale-of-redis-docker-and-distributed-problem-solving/): A multi-agent debugging experiment analyzing Redis pub/sub deadlocks, Docker container networking, and automated hypothesis testing. - [Sage Theme Migration: Building a WordPress Virtual Server Rack](https://eddykawira.com/wordpress-migration-no-documentation/): Modernizing legacy CMS architectures with Laravel Acorn, Vite asset compilation, and containerized deployment pipelines. ## Author & Engineering Profiles - [About Eddy Kawira](https://eddykawira.com/about/): Background, technical philosophy, and infrastructure engineering experience. - [Homelab Architecture & Specifications](https://eddykawira.com/lab/): Production specifications for hypervisor compute, network topologies, storage pools, and out-of-band management. - [Technical Portfolio & Case Studies](https://eddykawira.com/projects/): Production software deployments, automation agents, and client systems architecture. - [Professional Resume](https://eddykawira.com/resume/): Summary of engineering credentials, core proficiencies, and enterprise infrastructure track record.