International Journal of Computer
Trends and Technology

Research Article | Open Access | Download PDF

Volume 72 | Issue 2 | Year 2024 | Article Id. IJCTT-V72I2P102 | DOI : https://doi.org/10.14445/22312803/IJCTT-V72I2P102

MLOps Antipatterns and Mitigation Approaches

Ankit Virmani, Manoj Kuppam

Received Revised Accepted Published
15 Dec 2023 21 Jan 2024 05 Feb 2024 19 Feb 2024

Citation :

Ankit Virmani, Manoj Kuppam, "MLOps Antipatterns and Mitigation Approaches," International Journal of Computer Trends and Technology (IJCTT), vol. 72, no. 2, pp. 9-15, 2024. Crossref, https://doi.org/10.14445/22312803/ IJCTT-V72I2P102

Abstract

Deploying machine learning models for various analytics and data applications at an enterprise scale brings diverse challenges. This paper breaks down these challenges and details the critical MLOps antipatterns - the practices to avoid while deploying the machine learning models. Like design patterns formalize software engineering wisdom, antipatterns help us recognize and communicate problematic methodologies. Some of these antipatterns stem from technical errors, while others arise from a lack of understanding of the context of ML usage. This paper aims to facilitate better documentation, collaboration, and faster problem resolution by establishing a common language around these antipatterns. In addition to outlining the antipatterns, the study provides solutions and best practices and suggests a path to a more mature MLOps approach.

Keywords

MLOps, Python, Security, Data, Deployment, DevOps.

References

[1] Jaime Hampton, Half of AI Models Never Make It To Production: Gartner, Datanami.com, 2022. [Online]. Available: https://www.datanami.com/2022/08/22/half-of-ai-models-never-make-it-to-production-gartner/
[2] A. Kumar, S. Gupta, and P. Rai, “Explainable AI in Practice: From Academia to Industry,” Proceedings of the ACM India Joint International Conference on Data Science and Management of Data, pp. 17-26, 2021.
[3] D. Sculley et al., “Hidden Technical Debt in Machine Learning Systems,” NIPS'15: Proceedings of the 28th International Conference on Neural Information Processing Systems, vol. 2, pp. 2503-2511, 2015.
[Google Scholar] [Publisher Link]
[4] Larysa Visengeriyeva et al., MLOps Principles, Web.Archive.org. [Online]. Available: https://web.archive.org/web/20210620182858/https:/ml-ops.org/content/mlops-principles
[5] ML Model Security – Preventing the 6 Most Common Attacks, Excella, 2021. [Online]. Available: https://www.excella.com/insights/ml-model-security-preventing-the-6-most-common-attacks
[6] Vitomir Jovanović, Mlflow-Model-Monitoring, Dev Genius, 2022. [Online]. Available: https://blog.devgenius.io/mlflow-for-model-monitoring-cb8b2177b67a
[7] Everton Gomede, Shapley Additive Explanations: Unveiling the Black Box of Machine Learning, Medium, 2023. [Online]. Available: https://medium.com/@evertongomede/shapley-additive-explanations-unveiling-the-black-box-of-machine-learning-477ba01ffa07