Technical Articles & Applied Opinions
Engineering perspectives on wrapping probabilistic systems in deterministic controls, structuring search relevance, and maintaining data pipeline integrity.
Featured article
AI Agent Security Architecture: Prompt Injection, Permissions, and Tool Execution
An agent that reads untrusted content, holds private data and can send messages is one crafted document away from a breach. This article shows where the authorization decision has to sit, what to log, and a synthetic demonstration you can run.
· 10 min read
Owning the Black Box: How to Audit Systems When the Builders Are Gone
When the people who built a system are gone, the system still leaves evidence. This guide shows how to turn that evidence into a technical audit an owner can act on: an asset inventory, a dependency map, and ranked findings.
· 6 min read
Legacy System Modernization: What Moves, What Stays, What Stops
A file can be ported while the work it supported is still missing. Use a migration ledger to track scope, then verify behavior and data ownership before retiring the old system.
· 6 min read
Building Task Management Around the Work
A practical guide to building task management: define the outcome, organize the work, make ownership and dependencies visible, and decide what counts as done.
· 9 min read
Dashboard Design: State, Exceptions, and Operational Decision Surfaces
A dashboard is useful when it helps an operator immediately decide what needs intervention and what can wait. It fails when it turns raw telemetry into a wall of saturated gauges without source freshness, alarm suppression, or drill-down context.
· 9 min read
RAID Is a Failure Budget, Not a Backup Strategy
RAID is part of a reliable-storage design, not the design itself. The useful questions are what can fail, how the workload behaves, what performance matters, and how recovery will work when a disk eventually needs replacement.
· 6 min read
S1000D Data Module Codes: Structured Identity and Faceted Navigation
The richest metadata in an S1000D corpus is sitting inside the identifier, yet most conversion pipelines treat it as an opaque string. Decomposed into typed segments, a DMC provides an immediate navigation taxonomy, a faceted query model, and exact-identifier routing that would otherwise take months to reconstruct out of prose.
· 6 min read
State Machine Verification for Multi-Step AI Pipelines
Orchestrating multi-step AI agents with loose prompt loops leads to silent deadlocks, invalid state transitions, and unreproducible failures. Modeling agentic workflows as formally verified finite state machines gives engineers deterministic transition invariants, exhaustive error recovery paths, and auditable replayability.
· 7 min read
The S1000D Transform Loop: One Model, Two Component Trees
Modernizing S1000D technical publications is rarely a frontend framework problem. The real work is resolving a complex reference graph and dynamic applicability into a clean intermediate model. Once the publication model is sound, rendering into React, Angular, or feeding context to an AI agent is straightforward execution.
· 6 min read
Non-Determinism Was Always the Job of Engineers. Software Developers Just Got a Pass.
Software engineering treated deterministic logic as the norm because, for decades, it mostly was. AI removes that exemption by putting real, unresolvable nondeterminism in the critical path. The discipline this now requires is not new — it is the tolerance, margin, and statistical thinking every other engineering field has always practiced.
· 5 min read
Beyond the Hype: Navigating Nuance in Applied AI
Applied AI discussion is often polarized between magical expectation and wholesale dismissal. We advocate for a nuanced approach that focuses on task decomposition, operational constraints, and hybrid architectures.
· 2 min read
Reconciling Fragmented Data: Pipeline Observability
As datasets scale across disparate software ecosystems, silent pipeline failures and schema drift inevitably corrupt reports and downstream models. We explain why data contracts and strict observability are essential.
· 3 min read
Search Relevance Starts with the Taxonomy
Vector search and embeddings are frequently marketed as a cure-all for enterprise retrieval issues. However, search relevance is only as reliable as the underlying taxonomy. We discuss hybrid search and structural metadata.
· 3 min read
Deterministic Controls Around Nondeterministic Models
Large Language Models are probabilistic by design, making them unstable foundations for business logic. We outline the architectural patterns required to wrap nondeterministic AI models in reliable, deterministic controls.
· 5 min read