Arnaud Pouppeville Architect’s Notebook

Arnaud Pouppeville Architect’s Notebook

Sign in Subscribe

Natural-Language Search Over Image Libraries: RAG Architecture

Artificial Intelligence

Natural-Language Search Over Image Libraries: RAG Architecture

How to make thousands of images searchable in plain language without CLIP: RAG architecture using vision at ingest, hybrid search, and custom retrieval glue.

Solving AI Sprawl: Using Git Worktrees and ADRs to Govern Parallel Agents

Software Architecture

Solving AI Sprawl: Using Git Worktrees and ADRs to Govern Parallel Agents

How a five-phase AI-assisted SDLC pipeline (SpecKit, Git worktrees, /done, Claude Poll, /fix-pr) enforces architectural discipline at speed without the AI sprawl.

Two Kinds of Human-in-the-Loop — And Why LangGraph Needs Both

Software Engineering

Two Kinds of Human-in-the-Loop — And Why LangGraph Needs Both

Two kinds of HITL interrupts in LangGraph: information gaps (stuck, need data) vs authority gaps (stopped, need approval). Design for both; use interrupt(), never END as pause.

Agentic AI Governance: It Fails at System Assumptions

Software Engineering

Agentic AI Governance: It Fails at System Assumptions

Agentic AI governance for mission-critical software: a three-tier model, ADR boundaries, and why AI fails at system assumptions, not syntax.

How C4 Model + DDD Fixed My Team's Architecture Diagrams

Software Architecture

How C4 Model + DDD Fixed My Team's Architecture Diagrams

How C4 model and DDD fixed our architecture diagrams: shared vocabulary, diagrams as code, and the C3 trap. Practical workflow for agile teams.

Medallion Architecture in Azure Databricks: The Practical Guide

Data Engineering

Medallion Architecture in Azure Databricks: The Practical Guide

A practical guide to Medallion Architecture in Azure Databricks — what Bronze, Silver, and Gold layers do, what they cost, and when skipping a layer makes sense.

How Z-Order Cut My Databricks ETL Time in Half

Data Engineering

How Z-Order Cut My Databricks ETL Time in Half

Z-Order cut my Databricks ETL from 1m40s to 50s per tenant. Here's why partitioning failed and how Z-Order works like a clustered index for Delta Lake.