đź“Ł Share your perspective

Help shape this research 🚀

Your experience with AI/ML-enabled software development can directly strengthen the next iteration of this work. If you have a few minutes, please answer the survey and contribute evidence that matters.

Respond survey now!

SE4AI: Structured AI/ML Enabled Software Development Process

Author: Felipe Sonntag Manzoni (UFAM / IComp - Sphere Research Group)

Venue: ICSE-Companion ’26 (Doctoral Symposium), Rio de Janeiro, Brazil

Poster companion page Last updated:

TL;DR (1-minute)

🎧 1-Minute Audio Summary

Prefer reading? The TL;DR bellow is designed to be read in about a minute.

Abstract

Most organizations developing software that depends on AI/ML models often rely on ad hoc or unstructured development practices. In many cases, the model and the surrounding software are developed separately, leading to several problems—particularly regarding the quality of the model, the compatibility of the system in which the model is embedded, and the overall suitability of the resulting system to the problem domain. Although prior research has addressed specific aspects of this process and proposed improvements for isolated phases, a comprehensive, quality-oriented development framework is still lacking. This doctoral research aims to propose an enhanced development process that addresses these gaps and opportunities, ensuring both methodological rigor and the delivery of reliable, high-quality AI/ML-enabled systems.

1-Minute content resume

Problem: AI/ML-enabled systems are often built with ad hoc practices, where model development and surrounding software evolve separately—creating quality, integration, and reproducibility risks.

Proposal: SE4AI is a structured, quality-oriented development process for AI/ML-enabled software that integrates AI/ML-specific activities and explicit quality checkpoints across the lifecycle.

Method: The process is refined and validated through evidence-based research methods (e.g., SLR/grounded theory) and empirical evaluation (experiments and industrial case studies).

Expected impact: Better structure, traceability, and reliability—bridging the gap between experimentation and production-ready AI software.

Resources

📄 Paper (PDF) 🖼️ Poster (PDF) 📎 Presentation (PDF)

Full References

  1. Angelantonio Castelli, Georgios Christos Chouliaras, and Dmitri Goldenberg. 2025. Maturity Framework for Enhancing Machine Learning Quality. KDD ’25. doi:10.1145/3711896.3737246
  2. Rashina Hoda. 2022. Socio-Technical Grounded Theory for Software Engineering. IEEE TSE 48(10). doi:10.1109/TSE.2021.3106280
  3. Barbara Kitchenham et al. 2009. Systematic literature reviews in software engineering – A systematic literature review. IST 51(1). doi:10.1016/j.infsof.2008.09.009
  4. Giuliano Lorenzoni et al. 2021. Machine Learning Model Development from a Software Engineering Perspective: A Systematic Literature Review. arXiv:2102.07574
  5. Lucy Ellen Lwakatare et al. 2020. Large-scale machine learning systems in real-world industrial settings. IST 127:106368. doi:10.1016/j.infsof.2020.106368
  6. SĂ´mulo Mafra, Rafael Barcelos, and Guilherme Travassos. 2006. Aplicando uma Metodologia Baseada em EvidĂŞncia... SBES. doi:10.5753/sbes.2006.21216
  7. Silverio Martínez-Fernández et al. 2022. Software Engineering for AI-Based Systems: A Survey. TOSEM 31(2). doi:10.1145/3487043
  8. Elizamary Nascimento et al. 2020. Software engineering for artificial intelligence and machine learning software: A systematic literature review. arXiv:2011.03751
  9. Alex Serban et al. 2020. Adoption and Effects of Software Engineering Best Practices in Machine Learning. ESEM ’20. doi:10.1145/3382494.3410681
  10. Forrest Shull, Jeffrey Carver, and Guilherme H. Travassos. 2001. An empirical methodology for introducing software processes. ESEC/FSE-9. doi:10.1145/503209.503248
  11. Klaas-Jan Stol, Paul Ralph, and Brian Fitzgerald. 2016. Grounded theory in software engineering research: a critical review and guidelines. ICSE ’16. doi:10.1145/2884781.2884833

Cite this work (BibTeX)

@inproceedings{manzoni2026se4ai,
  title     = {SE4AI: Structured AI/ML Enabled Software Development Process},
  author    = {Felipe Sonntag Manzoni},
  booktitle = {2026 IEEE/ACM 48th International Conference on Software Engineering (ICSE-Companion '26)},
  year      = {2026},
  address   = {Rio de Janeiro, Brazil}
}
How about a espresso patronum? Support me on Ko-fi

Contact

Felipe Sonntag Manzoni
Federal University of Amazonas (UFAM) — Manaus, Brazil
Institute of Computing — ICOMP — Sphere Research Group
Email: fsm2@icomp.ufam.edu.br

Affiliations/Supporters

UFAM IComp SPHERE Research Group - PPGI-IComp