We design and implement custom-tailored AI systems for your business and your industry.
Integration with existing systems
AI systems are most often developed in connection to existing data sources. These could, for example, be ERP systems, industrial automation systems, databases with customer transactions and other data sets, which make up the data foundation for the AI. As part of the data importing process, we also handle import automation and data cleansing.
Our AI development process differs from project to project, but it generally follows the following procedural model that consists of four phases (see image):
- Problem analysis and definition
- Data collection
- Deployment and operation
We handle the entire process but are also happy to take over partial steps in collaboration with your IT teams.
- Deep Learning — TensorFlow, Keras
- Classical CPU-based ML — Sklearn
- Automation of integration and deployment — Kubeflow Pipelines
- Defence — IBM Adversarial Robustness 360 Toolbox
- Training — in-house GPU Hardware or cloud
- Integration with big data