I wrote a technical book on the infrastructure layer behind LLM applications - the part between "call the API" and "it works in production."

Story behind: Applying for jobs on LinkedIn and reading job requirements for AI companies. Asking Chatgpt about each topic and getting answers. . asked how about book about it. ChatGpt said great. Asked Claude code to help me with it from our conversations.

  Covers: distributed systems foundations, LLM gateways, RAG pipelines,                                                                                                                                       
  vector search, agent orchestration, multi-provider routing, cost                                                                                                                                            
  control, and system design walkthroughs.                                                                                                                                                                                                                                                                                                                                             
                                                                                                                                                                                                              
  GitHub: https://github.com/MandeepSinghthakur/production-ai-systems                                                                                                                                                                                    
                                                                                                                                                                                                              
  Written in TypeScript, runs on Node 22.6+ with zero dependencies -                                                                                                                                          
  no Docker, no API keys, no npm install. Clone and run.                                                                                                                                                      
                                                                                                                                                                                                              
  Target audience: engineers who know software but haven't run LLM                                                                                                                                            
  systems in production yet.                                                                                                                                                                                  
                                                                                                                                                                                                              
  Looking for technical review .