CoChat
Sign Up
Cited by 34Software Engineering Techniques and PracticesSoftware System Performance and ReliabilitySoftware Engineering Research

A Systematic Review of Machine Learning DevOps

Mboweni, Tsakani, Masombuka, Themba, Dongmo, Cyrille·2022·2022 International Conference on Electrical, Computer and Energy Technologies (ICECET)
DOI 10.1109/icecet55527.2022.9872968
Open this paper in CoChat — free account

Abstract

DevOps and Machine Learning (ML) on their own are insufficient in certain organisational context. To achieve the desired results, industries are now combining the benefits of DevOps automation and machine learning capabilities. This synergy is known as Machine Learning DevOps (ML-DevOps). However, there is currently no shared vision, strategy, or guidance across industry enterprises on how to use ML-DevOps for application development. While still in its infancy, ML-DevOps has showed some prospects to become a dynamic approach for developing and delivering secure, robust, and scalable applications. This article describes a systematic review of peer-reviewed conference papers and journal articles on ML-DevOps from 2015 to 2022 to ascertain the state-of-the-art and then identify any gaps in the current literature. Using empirical analysis, we systematically examined more than 60 studies that primarily deal with ML-DevOps (commonly referred to as MLOps). We combined meta analysis, document analysis and triangulation to map how ML-DevOps is currently understood and how it compares and differs from related approaches such as DevOps. The articles used were accessed via credible scholarly databases. The results of our search strategy retrieved articles that primarily discuss the topics of ML and DevOps synergy. The review of the existing literature shows a paucity of studies that strictly investigate ML-DevOps/MLOps. In most cases, the focus of these studies is on clarifying the meaning of MLOps. Our study discovered that although there are common elements in the definitions of MLOps, there remains a lack of standardisation on how it is defined, which sometimes cause some misconceptions and ambiguity. Finally, although there is significant literature written on MLOps, the review did not find evidence that there is a common understanding amongst scholars and experts on how MLOps should be implemented and institutionalized across industry to create a common vision.

Integrity signals

Registry

Crossref · 10.1109/icecet55527.2022.9872968

Cites

30 of 30 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Mboweni et al. [17] provides a higher-level overview of MLOps definitions, presents key roles of MLOps and associated tasks, and describes MLOps activities by constructing an MLOps architecture covering ML lifecycle.

    “The reported content provides a higher-level overview of MLOps definitions, presents key roles of MLOps and associated tasks, and describes MLOps activities by constructing an MLOps architecture covering ML lifecycle.”

    From A Multivocal Review of MLOps Practices, Challenges and Open Issues (2026) · cites this paper as [17]

  • Mboweni et al. [17] aimed to report how MLOps is understood and differs from DevOps.

    “Specifically, this work aimed to report how MLOps is understood and differs from DevOps.”

    From A Multivocal Review of MLOps Practices, Challenges and Open Issues (2026) · cites this paper as [17]

  • Mboweni et al. [17] reported an SLR on the state-of-the-art DevOps practices in ML-enabled systems.

    “Mboweni et al. [17] reported an SLR on the state-of-the-art DevOps practices in ML-enabled systems.”

    From A Multivocal Review of MLOps Practices, Challenges and Open Issues (2026) · cites this paper as [17]

  • The conceptualization of MLOps in earlier systematic studies aligns closely with our RQ1.1 results (e.g., [14], [17]).

    “Conceptualization: The conceptualization of MLOps in earlier systematic studies aligns closely with our RQ1.1 results (e.g., [14], [17]).”

    From A Multivocal Review of MLOps Practices, Challenges and Open Issues (2026) · cites this paper as [17]

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Mboweni, T., Masombuka, T., & Dongmo, C.
Affiliations
University of South Africa (ZA)
ORCID
Venue
2022 International Conference on Electrical, Computer and Energy Technologies (ICECET)
Publisher
IEEE
Published
2022-07-20
Pages
1-6
DOI
10.1109/icecet55527.2022.9872968
OpenAlex
W4297786875
Semantic Scholar
ceb83d54d7243f2d9fc2a6a77dba966ba1de1746
Cited by
34
References
29
Citations by year
2026: 5 · 2025: 12 · 2024: 12 · 2023: 5
FWCI
0.91
Type
journal article
Language
en
Open access
closed
Licence
Locations
International Conference on Electrical, Computer and Energy Technologies (ICECET) (publishedVersion)
Topics
  • Physical Sciences › Computer Science › Information Systems › Software Engineering Techniques and Practices
  • Physical Sciences › Computer Science › Computer Networks and Communications › Software System Performance and Reliability
  • Physical Sciences › Computer Science › Information Systems › Software Engineering Research
Keywords
MLOps, DevOps, machine learning deployment, software development automation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review