Stephen Alexander

Contact LinkedIn RSS

21 pieces · Newest first

Writing

Essays and articles on operational AI, distributed systems, and security. Subscribe via RSS.

Essays

These form a series. See how they relate →

The Inference Control Plane, Part One

The placement contract: where inference runs determines what it can know, what it can do, and whether it can be trusted.

Download PDF ↓

From SBOM to Decision BOM

How inventories create false confidence in AI decisions — a complete decision record is not the same as sufficient evidence. Companion to Evidence Debt.

Download PDF ↓

The system was designed to operate, not to prove what it did.

Download PDF ↓

Operational AI in Search and Rescue

One system, decomposed three ways: by function, by placement, and by authority.

Download PDF ↓

Operational AI Is a Distributed Systems Problem

As agentic workloads scale, durable value moves from components to composition. (Second Edition — the expanded foundational essay.)

Download PDF ↓

Whitepapers

Digital Transformation and IT Modernization for Elections in AWS

An AWS whitepaper on modernizing election systems for security and scale.

Classic Intrusion Analysis Frameworks for AWS Environments

Adapting traditional intrusion analysis frameworks to the cloud.

Articles

Are your APIs actually Agentic Ready?

Why APIs must be redesigned for autonomous AI agents as primary consumers.

Features are easy. Validation should be too.

Building a smoke-test system to catch what unit tests miss in AI-generated code.

Trust, but Verify: Adversarial Review for AI Code

Adversarial AI code review surfacing real findings that passed traditional testing.

Specs vs Vibes: The Difference Between Building and Hoping

Spec-driven development with AI agents versus vibe coding.

Building an AI Learning Hub

Building a knowledge platform with Claude Code — why hooks, not prompts, are the guardrails.

TIL: What I Learned by Studying Agentic AI in Practice

700+ controlled runs showing context, not just capability, determines agentic outcomes.

Building an Agent Workflow Observatory to Study Decision-Making in AI Agents

Observing how AI agents make decisions under controlled constraints.

Lessons from the Human-AI Partnership: Coding Beyond the Lines

Why effective human-AI partnerships depend on context, design, and treating AI as a teammate.

Starting with the End in Mind: How Amazon's Working Backwards Principle Transforms AI Development

Applying Amazon's Working Backwards method to AI projects.

Cloud Sayings Showdown: Which AI Model Comes Out on Top?

Comparing Claude and GPT models across response time, output, and user feedback.

From Idea to Code: What I Learned Using AI to Build Websites and APIs

Practical lessons from building with AI tools like Claude, Cursor, and v0.

AI is Eating the World: A Culinary Journey of AI Models and Approaches

Demystifying AI model types through restaurant analogies.

Did I build an LLM Gateway?: A Curious Developer's Journey

Accidentally building an LLM gateway while handling multiple foundation models.

My Journey into AI Security: What Every Cybersecurity Professional Should Know

Key AI security threats: adversarial ML, prompt injection, and AI-powered social engineering.