AI Engineering Research

An open index of evidence, theory, and practice

research-document

AI Engineering Research Program

Research Area
General Research
Status
draft
Version
0.1
Priority
medium
Updated
2026-07-22
Source
content/disciplines/ai-engineering/research-program.md

AI Engineering Research Program

Master Architecture and System of Record

Mission

Develop a scientifically grounded Theory of AI Engineering.

The goal is not to write a book.

The goal is to build a permanent body of engineering knowledge. Every other artifact is derived from that knowledge.

Primary System of Record

The Scientific Research Journal is the authoritative source for the project.

No idea exists unless it first exists in the journal.

Every other artifact is derived from it.

Scientific Research Journal
│
├── Evidence Registry
├── Hypothesis Registry
├── Theory Registry
├── Theory Evolution Log
├── Decision Framework Registry
├── Knowledge Base
├── Glossary
├── Concept Map
├── Principle Dependency Graph
├── Experiment Registry
├── Open Questions Registry
├── Pattern Library
├── Anti-pattern Library
└── Metrics Registry

Repository Layout

ai-engineering-research/
│
├── 00-governance/
│   ├── constitution.md
│   ├── research-method.md
│   ├── evidence-standards.md
│   ├── citation-policy.md
│   └── contribution-guide.md
│
├── 01-journal/
│   ├── journal/
│   └── index.md
│
├── 02-registries/
│   ├── evidence/
│   ├── hypotheses/
│   ├── theories/
│   ├── theory-evolution/
│   ├── experiments/
│   ├── decision-frameworks/
│   ├── glossary/
│   ├── metrics/
│   └── open-questions/
│
├── 03-knowledge/
│   ├── knowledge-base.md
│   ├── concept-map.md
│   ├── principle-dependency-graph.md
│   ├── engineering-principles.md
│   ├── design-patterns.md
│   └── anti-patterns.md
│
├── 04-research/
├── 05-experiments/
├── 06-theory/
├── 07-book/
├── 08-derived/
├── 09-tooling/
└── 10-archive/

Research Roadmap

  1. Build research infrastructure.
  2. Define AI Engineering as a discipline.
  3. Discover first principles.
  4. Research supporting sciences.
  5. Research AI technologies.
  6. Build AI Engineering theory.
  7. Catalog engineering patterns.
  8. Catalog engineering anti-patterns.
  9. Develop engineering economics.
  10. Research human factors.
  11. Build a unified theory.
  12. Attempt to falsify the theory.
  13. Produce engineering frameworks.
  14. Generate books, courses, and other derived artifacts.
  15. Continue continuous research.

Recommended Research Agent Teams

Core Theory

  • AI Engineering Theory
  • Theory Integration
  • First Principles
  • Scientific Methodology
  • Systems Thinking

Computer Science

  • Software Engineering
  • Computer Science Foundations
  • Systems Engineering
  • Distributed Systems
  • Information Theory
  • Control Theory
  • Probability & Statistics
  • Complexity Science

AI Science

  • Transformer Architecture
  • Large Language Models
  • Training
  • Inference
  • Attention
  • Reasoning
  • Retrieval-Augmented Generation
  • Memory
  • Tool Use
  • Agent Systems
  • Evaluation
  • Multimodal AI

AI Engineering

  • Context Engineering
  • Prompt Engineering
  • Workflow Engineering
  • Artifact Engineering
  • Memory Engineering
  • Verification Engineering
  • Evaluation Engineering
  • Reliability Engineering
  • Observability
  • Governance
  • Security
  • Operations

Human Sciences

  • Cognitive Psychology
  • Human-Computer Interaction
  • Learning Science
  • Decision Science
  • Organizational Psychology

Economics

  • Token Economics
  • Productivity
  • Return on Investment
  • Cost Modeling
  • Verification Economics

Knowledge Sciences

  • Knowledge Management
  • Information Architecture
  • Ontology Engineering
  • Knowledge Graphs
  • Semantic Search

Meta Research

  • Evidence Quality
  • Devil's Advocate
  • Replication
  • Gap Analysis
  • Terminology
  • Research Process Improvement

Standard Scientific Workflow

  1. Review the journal.
  2. Summarize current understanding.
  3. Identify the largest uncertainty.
  4. Generate research questions.
  5. Form hypotheses.
  6. Challenge assumptions.
  7. Gather evidence.
  8. Compare competing viewpoints.
  9. Search for counterexamples.
  10. Update the mental model.
  11. Append a journal entry.
  12. Update all registries.
  13. Select the next research objective.

Information Flow

Research
    ↓
Scientific Research Journal
    ↓
Evidence Registry
    ↓
Hypothesis Registry
    ↓
Theory Registry
    ↓
Engineering Principles
    ↓
Decision Frameworks
    ↓
Knowledge Base
    ↓
Books
Courses
Presentations
Training
Tools
Standards

Candidate First Principles (Research Targets)

These are hypotheses, not accepted theory.

  • Probabilistic Generation
  • Context Dependence
  • Externalized State
  • Verification over Trust
  • Controlled Delegation
  • Information Quality
  • Human-AI Collaboration
  • Cost-Constrained Computation
  • Measurable Engineering Outcomes

Each principle must survive repeated attempts at falsification before entering the Theory Registry.

Success Criteria

The research program succeeds when:

  • Every conclusion is traceable to evidence.
  • Competing viewpoints have been investigated.
  • Failed hypotheses are preserved.
  • Engineering principles explain multiple phenomena.
  • Decision frameworks emerge naturally from theory.
  • Educational artifacts are generated from the Scientific Research Journal rather than written independently.