-
Problem Formulation
- Clarifying questions
- Use case(s) and business goal
- Requirements
- Constraints
- Data: sources and availability
- Assumptions
- ML formulation
-
Metrics
- Offline metrics
- Online metrics
-
Architectural Components
- High level architecture
-
Data Collection and Preparation
- Data needs
- Data Sources
- Data storage
- ML Data types
- Labelling
-
Feature Engineering
- Feature selection
- Feature representation
- Feature preprocessing
-
Model Development and Offline Evaluation
- Model selection
- Dataset construction
- Model Training
- Model eval and HP tuning
- Iterations
-
Prediction Service
-
Online Testing and Deployment
- A/B Test
- Deployment and release
-
Scaling, Monitoring, and Updates
- Scaling (SW and ML systems)
- Monitoring
- Updates