Field notes

Production AI, written plainly.

Notes on the unglamorous parts of AI systems: data contracts, runtimes, review loops, compute limits, and the engineering habits that keep them useful after the demo.

The Agentic Trap: Why AI Systems Need Fewer Paths and More Truthful State

The fastest way to make production AI fail is to let execution paths disagree.

AI business systems

The Runtime Boundary Is the Product: Why AI Workflows Fail Outside the Model

Whether the surrounding runtime can make the model usable, observable, recoverable, and repeatable.

AI runtime architecture

The Critic Needs a Contract: Why AI Review Systems Need Profiles, Schemas, and Evidence

A useful AI reviewer needs versioned behavior, constrained output, evidence, and failure metadata.

AI review systems

From Human Taste to Training Data: Making High-Volume AI Output Reviewable

Human review only scales when the system can learn how to filter.

AI product learning loops

Scarce Compute Is a Product Constraint: Designing AI Workflows Around One GPU

A single GPU changes what can run together, what must wait, and what the system must admit.

AI product operations

Running Qwen3.6 27B INT4 on a 5090

A field note on local VLM critic runtime, quantization, startup, and structured output.

Local AI runtime notes

Benchmark Before You Optimize: Turning AI Generation Speed Into a Measurable Loop

AI generation speed becomes tractable when the team measures workflow phases.

AI workflow optimization

Prompt Tokens Are Product Budget: Why AI Image Prompts Need Measurement, Not Vibes

Prompt length is the budget available for product decisions and constraints.

AI image generation

Manifests Beat Memory: Making AI Generation Runs Resumable and Auditable

Long-running AI work should depend on manifests the system can inspect, resume, and verify.

AI operations

When Humans Can Move Files: Why AI Image Pipelines Need One Canonical Storage Model

The fix for ordinary storage failure is one clear source of truth.

AI data architecture

The Trigger Token Is a Contract: Avoiding Silent LoRA/Base Model Mixups

A LoRA run is a contract between model, weights, trigger token, runtime, and review expectations.

LoRA runtime design

When a LoRA Looks Worse: Debugging Training Quality Before Blaming the Model

Treat a bad LoRA output like an investigation, not a verdict.

LoRA evaluation

Do Not Train on Chaos: Why LoRA Datasets Need Copy-Only Data Prep

LoRA quality begins before training starts, with data prep the team can still explain.

LoRA training

The Remote Review Trap: When a Portal Becomes a Second Product

Remote review should be a narrow surface over canonical data, not a second product.

Human-in-the-loop AI