Every ML project begins with an initial model development phase that includes exploratory data analysis, feature engineering, model creation, training, and evaluation. During this phase, data science teams prioritize delivering a high-performing model, however, generating the ML models goes beyond this phase. While model development requires deep data science expertise, deploying the model demands operational skills.
The key challenges in moving ML model from development to production includes:
- Data issues
- Code issues
- Tracking of models
- Deployment issues, and
- Scalability issues
We propose a six-step method for ML Operationalization that can help overcome these challenges.
Download the whitepaper to read more.
Sandhya M NTechnology Lead, Data & AI Studio
Harshit ParikhSenior Vice President, Data & AI Studio
