Agile4MLS: Leveraging Agile Practices for Developing Machine Learning-Enabled Systems — An Industrial Experience
Authors
Karthik VaidhyanathanAnish ChandranHenry MucciniRegi Roy
An industrial experience report co-authored with a university software-architecture group.
Abstract
Increased adoption of machine learning (ML) in business has led to a surge in ML-enabled systems in the market giving rise to new challenges. This article provides insights into Agile4MLS, highlighting the challenges we faced and the process we followed.
IEEE Software abstracts are short by house style. This is the complete published abstract.
Why we worked on this
Machine learning breaks the assumptions agile development is built on. You cannot estimate a task whose outcome depends on whether the data supports it, and “done” stops being a binary when the deliverable is a model with a confidence interval. This paper is our own account of running real projects under that tension, published in IEEE Software instead of asserted on a website, which is why it sits underneath the claim, made elsewhere on this site, that AI is built into how we deliver, not bolted onto what we sell.
Where to start
The AI Modernization Assessment is two weeks and $5,000. Not a code audit. You get a prioritized roadmap, effort and cost ranges, and a written list of the workflows where AI would add risk without adding value.
Book the assessment