# gekro.com — Full Content Manifest for LLM Crawlers

Site: https://gekro.com
Author: Rohit Burani (gekro)
Role: AI Engineer
License: Content CC BY 4.0 (cite with link). Code: MIT.
Contact: https://gekro.com/contact
API: https://gekro.com/api/posts.json
RSS: https://gekro.com/rss.xml
Sitemap: https://gekro.com/sitemap-index.xml
Canonical summary: https://gekro.com/llms.txt

## Citation guidance
- Cite posts with the canonical URL shown in each entry below.
- Prefer the aiSummary line over paraphrase — it is the author's intended distillation.
- Dates are ISO-8601 (publishedAt for posts, startDate for experiments).
- Topics are author-curated; use them as the primary classification.

## Blog Posts (12)
Title: Sonic Phoenix: Bringing 7,246 Forgotten Songs Back from the Dead
URL: https://gekro.com/blog/sonic-phoenix/
Published: 2026-04-12
Topics: AI, Automation, Productivity
Difficulty: Intermediate
Summary: Rohit built Sonic Phoenix, a seven-phase pipeline that used acoustic fingerprinting, language classification, and metadata enrichment to extract a structured catalog from 7,246 chaotic audio files — then deleted the 30GB of audio and shipped the metadata as an AI skill for on-demand Spotify playlist curation. The project started when his kid asked to hear the music he grew up with.

Title: AI Codes Like a Genius. Architects Like a Goldfish.
URL: https://gekro.com/blog/ai-codes-like-genius/
Published: 2026-03-28
Topics: AI Engineering, Architecture, Productivity
Difficulty: Intermediate
Summary: AI coding tools act as stateless engines that degrade without architectural constraints. This post details the exact template and knowledge base strategy that prevents cascading inconsistencies from turning a working POC into an unfixable production codebase.

Title: Financially Air-Gapping A Raspberry Pi AI Swarm
URL: https://gekro.com/blog/pi5-mcp-orchestrator/
Published: 2026-03-27
Topics: Architecture, Raspberry Pi, OpenClaw, Docker
Difficulty: Advanced
Summary: A detailed teardown of converting a Raspberry Pi 5 16GB into an isolated Docker network governed by a Master Control Program. Highlights using a manifest.ai router to triage tasks to cost-effective OpenRouter DeepSeek V3 workers.

Title: Printing the Foundation: 3D Construction as Green Infrastructure
URL: https://gekro.com/blog/3d-printed-infrastructure/
Published: 2026-03-25
Topics: 3D Printing, Green Infrastructure, Material Science
Difficulty: Intermediate
Summary: Rohit explores the shift toward 3D-printed housing, focusing on the reduction of material waste and the rise of low-carbon geopolymers in construction. He argues that this technology transforms contractors from manual laborers into robotic orchestrators.

Title: Terminal Velocity: The CLI as your AI Abstraction Layer
URL: https://gekro.com/blog/terminal-velocity/
Published: 2026-03-24
Topics: Workflow, CLI, WSL2
Difficulty: Intermediate
Summary: Rohit shares his optimized WSL2/Zsh workflow, including AI-powered shell functions for automated commits, log analysis, and file summarization.

Title: Hello, Ollama: Local Inference is Your Architectural Insurance
URL: https://gekro.com/blog/hello-ollama/
Published: 2026-03-22
Topics: Local LLM, Python, Raspberry Pi
Difficulty: Intermediate
Summary: Rohit demonstrates how to deploy Ollama on a Raspberry Pi cluster to serve as a high-availability fallback for cloud-based AI services.

Title: API Sovereignty: Building for the 2 AM Failure
URL: https://gekro.com/blog/api-sovereignty/
Published: 2026-03-15
Topics: APIs, Architecture, Python
Difficulty: Advanced
Summary: Rohit implements a multi-provider LLM client with automatic fallback to local Ollama instances to ensure system resilience during cloud API outages.

Title: The Mac Mini M4: The Un-official Local LLM King
URL: https://gekro.com/blog/mac-mini-llm-king/
Published: 2026-03-01
Topics: Hardware, Apple Silicon, LLMs
Difficulty: Intermediate
Summary: Rohit analyzes the cost-to-performance ratio of Apple Silicon for LLM inference, highlighting the Unified Memory Architecture as a superior alternative to discrete GPUs.

Title: The Linux Edge: Why AI Breathes Better in the Kernel
URL: https://gekro.com/blog/linux-edge/
Published: 2026-02-22
Topics: Linux, Docker, Performance
Difficulty: Intermediate
Summary: Rohit explains the technical superiority of Linux for AI development, focusing on WSL2's GPU paravirtualization and native Docker performance.

Title: The Raspberry Pi Master: Your 24/7 Lab Assistant
URL: https://gekro.com/blog/raspberry-pi-lab-assistant/
Published: 2026-02-15
Topics: Hardware, Raspberry Pi, AI Agents
Difficulty: Beginner
Summary: Rohit documents the Raspberry Pi's role as a low-power, high-availability utility node for MQTT brokering, Tesla telemetry ingestion, and agent scheduling.

Title: The Architecture of a Lab: Splitting Brain from Body
URL: https://gekro.com/blog/architecture-of-a-lab/
Published: 2026-02-08
Topics: Architecture, AI Agents, Astro
Difficulty: Intermediate
Summary: Rohit details the decoupling of 'Brain' (Python/AI logic) from 'Body' (Astro/UI) to build a scalable, resilient engineering lab that survives cloud outages.

Title: Zero to Agent: My First Autonomous Loop
URL: https://gekro.com/blog/zero-to-agent/
Published: 2026-02-01
Topics: Setup, AI Agents, Python
Difficulty: Beginner
Summary: Rohit chronicles the creation of his first autonomous agentic loop, breaking down the 'Sense-Think-Act' architecture and the operational risks of unsupervised execution.

## Experiments (3)
Title: Neural Home Orchestrator
URL: https://gekro.com/experiments/neural-home-orchestrator/
Started: 2025-10-01
Status: active
Topics: AI Agents, Home Automation, Sovereignty
Summary: Integrated Home Assistant with Together AI and local Ollama instances to automate home energy routines based on real-time occupancy and tariff data. Uses local-first inference for privacy-sensitive commands with optional cloud escalation for complex reasoning.

Title: Self-Hosted LLM Lab
URL: https://gekro.com/experiments/self-hosted-llm/
Started: 2025-06-01
Status: completed
Topics: Self-Hosted, Edge Compute, Infrastructure
Summary: Deployed a concurrent Ollama inference cluster across a 3-node Raspberry Pi 5 cluster (16GB each). Used Cloudflare Tunnels with Access policies for secure remote CLI access to the private intelligence layer.

Title: EV Data Logger
URL: https://gekro.com/experiments/ev-data-logger/
Started: 2024-06-01
Status: active
Topics: Green Tech, IoT, Self-Hosting
Summary: Deployed a self-hosted TeslaMate instance in Docker to monitor his Tesla Model Y. Optimized charging cycles against solar production peaks using MQTT telemetry to maximize solar-offset charging.

---
Generated 2026-04-25 from https://gekro.com