How people, repositories, knowledge systems, evaluation frameworks and AI work together to improve engineering, research, creativity and decision-making.
View the Project on GitHub dermot-r-cochran/engineering-intelligence
Every page in this knowledge base states a principle. This page is the evidence: the repositories, by the same author, where each principle is already applied, tested or found wanting. A principle with no repository behind it is a hypothesis and is marked as one.
| Topic | Practised in | What to look at |
|---|---|---|
| Repository-aware development | local-agent (private) | A coding agent that may reach GitHub, Azure DevOps, Azure and the local desktop and nothing else, with every write confirmed and every remote write in an actions ledger. Its decisions file records each ruling in “do not re-propose” form, which is a repository used as a knowledge system. |
| AI evaluation | applied-statistics-for-AI-engineers | A tutorial on the statistics an evaluation needs before a score means anything: variance, sample size, overdiagnosis, the difference between a metric and a decision. |
| Context engineering | careful-memory | A long-term memory for agents with Bayesian confidence, decay and explicit contradiction tracking: context that earns its beliefs rather than accumulating them. |
| Human-AI systems | world-model, shadow-architect | Durable, governed state for business AI systems; and adversarial testing of AI-generated architecture against system-level guardrails. Both put the human decision and its audit trail before the model. |
| Engineering intelligence | architecture-definition-model | A layered framework for stating an architecture so that decisions, constraints and evidence sit where a later engineer can find them. |
| Natural language programming | local-agent (private) | Intent in plain language, turned into confirmed file writes and commands with the diff shown first. The repository’s own finding, from use: natural language is the specification, the confirmation prompt is the verification. |
| Virtual anthropology | virtual-anthropology | The Archipelago, a deterministic, event-sourced virtual civilisation built as a research platform for exactly the questions that page asks. |
| Knowledge representation (no page yet) | foundation-model, swarm, visual-llm | Epistemic foundation-model architectures for representing beliefs and experience, and a hierarchical image description language that puts a structured layer between raw images and decisions. Candidates for a page once their findings are written up. |