Systems

Practical builds, experiments, and architecture notes behind the writing.

These systems turn ideas from the articles into concrete workflows, demos, and engineering decisions.

These are not just projects.

They are explorations of how software systems behave when ideas meet constraints - scale, uncertainty, AI integration, and real-world complexity.

Systems writing AI-assisted engineering series How I built an AI engineering system

Explore by System Type

Use these paths to connect practical builds back to the writing themes.

AI-assisted engineering

Workflows for planning, coding, review, and orchestration

Learning and workflow experiments

Structured learning, document workflows, and evidence-rich AI outputs

Platform infrastructure

Small systems for API design, serverless architecture, and delivery discipline

Learning and workflow experiment

AI Workflow Lab

Turning everyday business documents into structured, reviewable data with evidence, confidence, and missing-field reporting.

Problem: Business documents are hard to trust when AI extraction does not show evidence.
Built: Document extraction demos with fields, confidence, evidence, and missing-field reporting.
Patterns: Evidence-first AI output, human review, structured extraction, workflow readiness.

Connects to writing about context, review, and making AI-assisted output inspectable.

View system Related writing: How Modern LLM Systems Really Work

Document -> Fields -> Evidence

AI-assisted engineering system

AI Dev Orchestrator

Designing an AI-assisted software engineering workflow across planning, coding, review, and deployment.

Problem: AI-assisted development needs orchestration, review loops, and human decision points.
Built: A workflow model for using multiple AI tools across planning, implementation, review, and merge discipline.
Patterns: Agent roles, issue-first delivery, review gates, evidence trails, human ownership.

Acts as practical evidence for the AI-Assisted Software Engineering series.

View system Related writing: How I Used Two AIs to Build a Software Engineering System Related series: AI-Assisted Software Engineering

Problem -> Approach -> What changed

Platform infrastructure system

Survey / Poll Serverless System

A focused serverless system to test API design, event flow, and deployment discipline using AWS.

Problem: Architecture ideas need small, concrete systems before they become reusable practice.
Built: A serverless API and persistence experiment for survey and poll workflows.
Patterns: API boundaries, event flow, Lambda execution, DynamoDB persistence, deployment discipline.

Shows the infrastructure side of building small systems to validate engineering decisions.

View system Related writing: Understanding the AI Ecosystem

Problem -> Approach -> What changed

Learning and workflow experiment

AI-native Learning Platform

Exploring how structured content and AI can support deeper technical learning.

Problem: Technical learning often optimizes for volume instead of structure, depth, and feedback.
Built: An early exploration of structured content, retrieval, and AI assistance for deeper learning.
Patterns: Learning paths, content structure, AI guidance, retrieval-oriented knowledge design.

Connects the site's AI foundations writing to experiments in learning systems.

View system Related writing: AI, Machine Learning and Data Science - Explained Simply

Problem -> Approach -> What changed

Systems are the evidence layer for the writing.

The articles explain the mental models. These systems test those ideas against practical constraints, interfaces, workflows, and review needs.

Read systems writing Start the AI engineering series