Pragmatic Programming in the Age of AI Coding Agents: An Empirical Case Study of Software Engineering Practices in AI-Assisted Development

---

📌 Executive Summary

This repository hosts the primary research workspace, data collection telemetry, BibTeX bibliography, and manuscript source for the empirical software engineering study:

"Pragmatic Programming in the Age of AI Coding Agents: An Empirical Case Study of Software Engineering Practices in AI-Assisted Development"

• Author: Tarunya Kesharwani
• Affiliation: 2nd Year B.Tech, Newton School of Technology
• GitHub: @tarunyaprogrammer
• Email: tarunyak.10@gmail.com
• Manuscript Format: Traditional 2-Column Academic Research Article

---

🔬 Empirical Research Framework & Methodological Principles

This study investigates how classical software engineering principles—specifically orthogonality, duplication avoidance (DRY), modularity, and test-driven guardrails—manifest when software implementation is partially or fully delegated to autonomous AI coding agents.

Scientific Principles Guiding This Study

• RQ-Driven Investigation: All metrics, repository mining procedures, and statistical tests map directly to explicit Research Questions.
• Evidence vs. Interpretation: Clear operational distinction between observed commit diffs, statistical correlations, analytical interpretations, and hypotheses.
• No Unsupported Causal Claims: Strict usage of correlational terminology (associated with, correlated with) to prevent unwarranted causal assertions.
• Reproducible Data Lineage: Every quantitative table value and plot is traceable through a transparent analysis pipeline:

$$\text{data/raw/} \longrightarrow \text{data/processed/} \longrightarrow \text{analysis/} \longrightarrow \text{Paper/main.pdf}$$

---

🎯 Research Questions (RQs)

• RQ1 (Principle Manifestation): How do traditional pragmatic software-engineering principles (e.g., orthogonality, DRY, modularity) manifest in AI-assisted software development?
• RQ2 (Engineering Practice Impact): How does AI-assisted code generation affect core practices such as refactoring, test co-evolution, and change propagation?
• RQ3 (Critical Guardrails): Which traditional engineering practices become most critical to system stability when implementation is delegated to AI agents?
• RQ4 (Emergent Defect Taxonomy): What specific categories of architectural friction and integration defects emerge in AI-generated code bases?

---

🗺️ Research Workspace Architecture (research/)

• 01-research-question.md: Formalized RQ specifications.
• 02-literature-review.md: Primary literature matrix & research gap identification.
• 03-pragmatic-principles.md: Operationalized principles into observable software engineering metrics.
• 04-methodology.md: Empirical repository mining methodology & commit classification.
• 05-research-log.md: Chronological research decision log.

---

📂 Public Repository Layout

ResearchPaper-Pragmatic_Programming_in_AI_World/
├── research/
│   ├── 01-research-question.md    # Formalized Research Questions (RQs)
│   ├── 02-literature-review.md    # Verified literature matrix & research gap
│   ├── 03-pragmatic-principles.md # Pragmatic principles mapped to SE metrics
│   ├── 04-methodology.md          # Empirical dataset & study methodology
│   └── 05-research-log.md         # Chronological research decision log
├── data/
│   ├── raw/                       # Raw git commit metadata & telemetry logs
│   └── processed/                 # Cleaned classified commit datasets (AI vs Human)
├── analysis/
│   └── extract_metrics.py         # Automated Python metric extraction script
├── Paper/
│   ├── main.tex                  # 2-column LaTeX manuscript source
│   ├── references.bib            # BibTeX bibliography database
│   ├── main.pdf                 # Compiled 2-column PDF manuscript
│   └── .gitignore               # LaTeX build exclusions
├── README.md                     # Project documentation (this file)
└── LICENSE                       # All Rights Reserved strict license

---

🛠️ Building & Previewing the Paper

Command Line Build

export PATH="/Library/TeX/texbin:$PATH"
cd Paper
pdflatex main.tex
bibtex main
pdflatex main.tex
pdflatex main.tex

---

📖 Citation Format

@article{kesharwani2026pragmatic,
  title     = {Pragmatic Programming in the Age of AI Coding Agents: An Empirical Case Study of Software Engineering Practices in AI-Assisted Development},
  author    = {Kesharwani, Tarunya},
  journal   = {Research Paper Series in AI-Assisted Software Engineering},
  year      = {2026},
  url       = {https://github.com/TarunyaProgrammer/ResearchPaper-Pragmatic_Programming_in_AI_World}
}

---

📜 License

Copyright (c) 2026 Tarunya Kesharwani. All rights reserved. See the LICENSE file for strict usage permissions.