A major goal is to understand how to prepare and plan the implementation of your first AI Project.
Description of the AI Plan Workshop
At the end of this 1-2 week workshop we concretise the outcomes of the Discovery Workshop and head from data exploration towards either a minimal viable model or recommendations towards addressing the bottlenecks.
If the data allows it, we want to move towards a minimal viable product, otherwise we deliver recommendations to address the bottlenecks and plan the next steps. In addition, we train employees to handle the state of art adoptions.
The Workshop start with exploring the data to assess the possibilities of the data assets.
- Assessing data quality and quantity
- Validate feasibility and usefulness of data for intended goals
- Guidance towards improved data
Selection of a suitable ML/AI model
Study the state-of-the-art approaches and choose an algorithm / model to apply to the selected use-case.
- Research for suitable models
- Select the appropriate open-source framework / implementation
- Define the scope of the MVP (minimal viable product)
- Identify relevant features and if applicable recommend new features
Initial model training & evaluation
First training of the model with the selected customer data
- Split the data into 3 subsets: training, validation, test.
- Launch and monitor training
- Evaluate and visualize results
- Recommendations to improve the model (performance metrics)
- Recommandations to integrate the model with your existing tools
- Identify outsourcing/partner selection criteria