Blog
Practical notes on forward deployed engineering, production readiness, and the decisions that make AI workflows usable.
-
Read the article
LLMs and System 2: Where Does the Checking Happen?
An opinion on LLM reasoning and the System 1/System 2 analogy, using one queue decision to ask where assumptions meet evidence.
-
Read the article
How to Unslop AI-Generated Code Before Your Next Release
Repair an AI-built app by reproducing one failing workflow, checking the fix, and deciding what to keep or replace. Includes a runnable retry example.
-
Read the article
AI Agent Harnesses: Stop Failed Runs and Hand Off the Evidence
Bound an agent run, stop repeated failures, and hand off useful evidence with a small JavaScript example you can run locally.
-
Read the article
One Shared IR for Python and JavaScript Expressions
Lower Python and JavaScript expressions to one common intermediate representation, with AST examples, validation, an interpreter, and a small JavaScript emitter.
-
Read the article
How to Prepare for an AI Engineer Interview: Build, Break, Explain
Prepare for an applied AI engineering interview by building one small agent, breaking it deliberately, and learning to explain its architecture, evaluation, safety, cost, and trade-offs.
-
Read the article
How I Use Coding Agents Without Giving Up Architectural Control
How I use repository law, bounded tasks, executable checks, evidence, and review to let coding agents work quickly without making architecture by accident.
-
Read the article
Forward Deployed Engineer vs Solutions Engineer vs Consultant
A practical comparison of forward deployed engineers, solutions engineers, and consultants for teams with a blocked AI deployment.
-
Read the article
AI Demo to Production: Readiness Checklist
Move an AI demo toward production with an illustrative, practical checklist for evaluation, security, failure handling, cost, ownership, and staged rollout.