AI Engineering Knowledge Base
From architecture choices to production troubleshooting, this hub organizes real AI Agent, MCP, LangGraph, Workflow and on-device AI problems. Each topic aims to provide reproducible methods, constraints and a concrete next step.
Engineering paths
01Design an AI AgentDecide whether you need an agent, then move through architecture, tools, memory, evaluation and production governance.02Connect models to real toolsChoose between function calling and MCP, then handle servers, transport, OAuth and permissions.03Build a recoverable agent workflowUse explicit state, checkpoints, pause/resume, concurrency and failure recovery when a simple chain is not enough.04Run deterministic automation reliablyFor fixed business processes, focus on webhooks, queues, retries, idempotency and observability.
Supporting systems
Recent engineering work
Gemini 4 Argon: Capabilities, Pricing, Migration, and a Real-Project Evaluation Plan10/14/2026EmbeddingGemma 2 Local Retrieval Test: Mac 270M, PDF/DOCX, 256d vs 768d10/10/2026Personal AI Agent Architecture: Memory, Permissions, App Actions and Local-vs-Cloud Design10/06/2026How to Adapt a UIKit App for iPhone Duo: Tab Bars, Size Classes, Safe Areas, and Reserved Regions09/28/2026AI Agent Sandbox Design: Hosted vs Self-Hosted, Files, Secrets, Network, and Persistence09/26/2026Transformers GGUF “Dequantizing the whole model” Fix: PyTorch Version Test on Apple Silicon09/26/2026
