Essays
The Journal253 essays
Long-form writing on AI, philosophy, psychology, and systems thinking.
-
Designing Data Pipelines for Machine Consumers
AI agents now consume more analytical data than humans at a majority of the organizations I work with. Machine consumers require fundamentally different quality contracts.
-
When Best Practices Become Worst Practices
Practices are context-dependent tools, not universal rules. Across the organizations I tracked, some standard practices harmed adopters in other contexts.
-
The Data Engineer’s Guide to Cost-Aware Architecture
Cost-aware architecture patterns cut monthly cloud data spend noticeably without degrading query performance. Five techniques every data engineer should apply.
-
The Data Analyst Role Is Being Redefined by AI
LLMs now handle routine SQL with high accuracy. Analyst job postings now ask for more domain expertise and less SQL. The role is being redefined.
-
AI in Research Has a Reproducibility Problem Ethics Frameworks Ignore
In the AI-assisted research papers I reviewed, most treated model outputs as findings and lacked reproducibility documentation. AI amplifies the replication crisis when ethics frameworks ignore methodology.
-
Autonomous Agents Need Ethical Guardrails, Not Ethical Training
Agents with prompt-based ethics still cross boundaries at scale. Architectural guardrails, not training, stop the violations that slip past instructions.
-
Why AI systems fail when humans don’t: The gap between statistical and experiential knowledge
-
Explainability Is Not Optional for Systems That Affect Lives
Explanation is a moral obligation when systems decide jobs, credit, or healthcare. Adding explainability cost little next to the harm of opaque decisions.
-
The Monorepo as Organizational Philosophy
Monorepos speed cross-team integration and reduce dependency failures, but they raise CI costs. The repository structure is an organizational chart in code.
-
State Management Is the Hardest Problem in Distributed Systems
Most distributed system failures I investigated traced to state management: split-brain conditions, stale caches, and concurrent update conflicts. State is where distributed systems break.