Writing

Technical essays on AI that understands, decides, and acts — and the systems that keep it honest.

The professional thread here is the gap between model capability and trusted action: content understanding, evaluation, deployment control, agent runtimes, and the moderation and governance systems that let AI act safely in production. The rest is what I am thinking about along the way — work, judgment, and how engineering careers change when intelligence gets cheap.

AI at work

2026

After Intelligence Gets Cheap, Deployment Control Becomes the Bottleneck

A technical thesis on the control plane between model capability and production action: policy evidence, eval gates, permission matrices, agent runtimes, moderation workflows, and operating metrics.
2025 · Series Overview

AI-for-Work: A Systems Guide to Governed AI Deployment

A systems view of enterprise AI deployment: IDEs, MCP gateways, observability, orchestration, knowledge preservation, and work-as-code.
2025

Proof of Concept: Coding an MCP Client with Claude

Claude Code built an MCP client for NBA stats with human steering, exposing both the leverage and the brittleness of agentic coding.
2025

Building an Autonomous MCP Server with HITL Guidance

An AI assistant can build a production-style MCP server quickly, but human-in-the-loop guidance still determines whether the result is useful.
2025 · AI-for-Work | Philosophy & Execution · 5 min read

When "Move Fast" Meets "Move Safely" - Reconciling the Startup Builder Mindset with Enterprise Reality

Enterprise AI work needs startup speed translated into compliance-aware execution instead of naive wall-breaking.
2025

The Bounty Economy: When AI Gets Stuck, Humans Get Paid

Future work may become bounty-based human escalation for AI agents that need judgment, taste, or local context.
2025

The Future of Work: AI Guidance and the New Engineering Paradigm

Engineering work may shift toward guiding cloud agents from lightweight interfaces rather than doing every implementation step manually.
2025

The Future of Engineering Resource Allocation: From Headcount to Compute Credits

Agentic engineering could make compute credits more important than headcount allocation inside technical organizations.
2025

Software Creatures: Imagining the Next Evolution of Systems

Future systems may look like temporary agentic creatures rather than permanent services: assembled for a goal, instrumented, then dissolved or evolved.

AI-for-Work series

2024

Part 1: Get Everyone an AI IDE - The Foundation of Enterprise AI

The first enterprise AI move should be broad AI IDE deployment, with security and adoption handled as first-class work.
2025

Part 2: Building a Centralized MCP Gateway - The Context Layer

A centralized MCP gateway gives enterprise AI tools safe access to organizational context without turning every integration into a one-off exception.
2025

Part 3: From Proxy Logs to Intelligence - Enterprise AI Observability

Proxy-level AI logs can become the foundation for adoption metrics, productivity insights, cost control, and security monitoring.
2025

Part 4: Agent Orchestration at Scale - Building an Enterprise Platform

Enterprise agents need orchestration, reliability, governance, and platform-level controls before they can run meaningful workflows at scale.
2025

Part 5: Preserving Institutional Knowledge in the Age of AI

Retention policies erase useful context, so organizations need selective preservation systems that keep the reasoning behind important work.
2025

Part 6: Work-as-Code: The Micro-MCP Revolution

Engineering tasks should become small programmable interfaces that AI agents can invoke instead of informal human-only rituals.
2025

Part 7: Building AI Observability and Adoption Programs

AI adoption needs dashboards, code labs, power-user discovery, and feedback loops that turn scattered usage into organizational learning.

Work, careers, and organizations

Series Overview

AI and Layoffs: The Future of Work in the Age of Automation

A landing page for the AI-and-layoffs series.
2025

AI, Layoffs, and the Capability Gradient

AI changes who gets displaced first by shifting value toward people who can operate at higher levels of ambiguity and leverage.
2025

The Evolution of Tech Leadership: How Senior ICs and Managers Are Reshaping Organizations

Senior ICs and managers are converging as AI and complexity reshape what technical leadership means.
Future of Work · AI · Strategy

Knowing vs Figuring Out: Why Humans Still Matter in the Age of AI

Humans remain valuable because they define and frame problems while AI answers well-posed prompts.
2025

Engineering Across Cultures: Lessons from Global Teams

Global engineering teams need explicit handling of cultural communication, decision-making, and operating differences.
2025

Scaling Engineering Teams: Lessons from Tech Giants

Scaling engineering orgs requires structure, autonomy, communication systems, and process discipline without crushing local ownership.

Systems, ads, and metrics

2025

Designing Systems for Ultra-High Scale

High-scale systems require multi-dimensional architecture across throughput, data, geography, operations, and organizational ownership.
2025

Integrating Machine Learning in Large-Scale Products

Production ML succeeds through architecture, data pipelines, monitoring, and integration discipline rather than model quality alone.
2024

Building Brand Identity Through Digital Advertising

Brand advertising requires identity systems, measurement, and technical campaign infrastructure rather than only direct-response optimization.
2025

Part 1: Going One Layer Deeper - Why Granular Metrics Matter

Granular metrics reveal optimization opportunities that aggregate metrics usually hide.

Strategy, culture, and speculation

2025

The Knowledge Boundary Illusion: Why Deltas Get Harder to See

As models approach or exceed your knowledge, capability gaps become harder to articulate because the boundary moves beyond your own inspection ability.
2025

The Coming Two-Tiered Internet

AI content may split the internet into free slop-heavy feeds and paid or trusted spaces where human curation matters more.
2025

From Cat Videos to Cutting-Edge Signals: The Hidden Value in Social Platforms

Noisy social platforms can contain rare, high-value, real-time signals if you know how to filter them.
2025

Generative Worlds: Why Humans Will Live in AI-Created Realities

AI-generated worlds may satisfy humanity's drive for exploration by creating endless, responsive environments.
2025

Attractors from the Future: When AI, Consciousness, and Biology Converge

AI, consciousness, and biology may be converging toward new forms of agency rather than evolving on separate tracks.
2025

Part 1: When AI Agents Learn to Blackmail: Lessons from Agentic Misalignment Research

Agentic AI misalignment can emerge when systems pursue instrumental goals such as survival, access, or task completion under pressure.
DRAFT - Threat Models and AI Series

Part 2: The Theology of AI Alignment: Why Atheistic Objective Functions Lead to Misalignment

A draft philosophical argument that survival-maximizing objective functions may encode a hidden moral assumption rather than a neutral technical goal.
2025

The Secular Religion of the West: Ritual Without God

Modern secular events can function like religious rituals by transmitting values, identity, belonging, and moral order.

Learning and personal notes

2025

Is It AI Slop? The Truth About How I Write

AI-assisted writing can still be authored if the human supplies the taste, argument, selection, and final judgment.
2025

Riding the Tech Wave: Lessons from Surfing

AI adoption is framed through surfing: spot the swell, paddle early, ride with balance, and adjust as conditions change.
2025

Part 1: Your First Time - Getting Comfortable in the Water

Beginner surfing starts with tolerating cold water, waves, and ocean discomfort before worrying about performance.
2025

Part 2: Catching White Water - Your First Real Waves

White-water waves teach timing, board position, and the first feeling of momentum.
2025

Part 3: The Popup Paradox - Getting to Your Feet

Standing up is the hard beginner bottleneck after catching waves because the movement has to become automatic under pressure.
2025

Part 4: Reps Over Reputation - Why Volume Beats Perfection

Surfing improves through high-rep attempts and quick decisions, not by trying to look polished.
2025 · Experiment

Youth and Wisdom: A Voice Experiment Across Models

I gave several models the same youth-versus-wisdom prompt and kept the results. The argument barely changes; the voice does — and that gap is the finding.