> ML_LITERATURE // SCULLEY-2015-HIDDEN-TECHNICAL-DEBT-IN-MACHINE-LEARNING-SYSTEMS_v1.0
Hidden Technical Debt in Machine Learning Systems
D. Sculley, Gary Holt, Daniel Golovin, Eugene Davydov, Todd Phillips, Dietmar Ebner, Vinay Chaudhary, Michael Young, Jean-François Crespo, Dan Dennison · Advances in Neural Information Processing Systems (NeurIPS) (2015)
mlops-production2015industry-standardnotAssessed
Principal Contribution
Identified systemic ML maintenance traps: glue code, pipeline jungles, data dependencies, configuration debt, and feedback loops, showing model code is only a tiny fraction of real systems.
Operational Relevance
Serves as qualified reference for implementing task-pipeline-orchestration in production systems.
Assumptions
- Underlying computational topology and mathematical bounds adhere to established convexity/smoothness guarantees
Limitations
- Hardware runtime speedups, privacy budgets, and convergence depend on hyperparameters and network communication limits
Connected Algorithms, Architectures & Tools
Related Algorithms:
Related Architectures:
Implementing Libraries: