Data science is a fast growing domain. As the ecosystem of software frameworks and tools is developing at a fast pace, it is hard to keep up. In contrast to software development, no standard workflow has been established yet. Most companies focus on building the right thing and less on building the thing right. A lot of data science projects fail, not because of missing data science know how, but because of a lack of a best practice workflow.
We introduce a framework that facilitates the collaboration within the data science team, ensures documentation and reproducibility of results, and makes the deployment of AI models child’s play.
Data-driven products require more than a well trained AI model. They consist of a whole pipeline from data ingestion to an output that generates insights or triggers actions. We have acquired a lot of experience in building end-to-end solutions on the Google Cloud Platform. Solutions that require low to no operations, scale without limits, and leverage the best of Google’s cutting edge technology.
Together with our strategic partner Panter, we deliver everything from the architecture to the implementation.
Developing data-driven products requires state of the art tools and a high-end infrastructure.
As a cloud-native company, we have been using the Google Cloud Platform (GCP) from the beginning. We are convinced that it has the best offering when it comes to data analytics and machine learning. We build on this cutting edge infrastructure on our own projects and when delivering our services to our clients.
With TensorFlow, Google's AI Hub, and the AI Platform on the GCP, we use the best in class resources and infrastructure to develop and run our prediction models. In addition to that, we use Kubeflow to cover the whole data science workflow of exploring, training at scale, and deploying.
by Juri Sarbach, CTO – published on Medium.com on 6 October 2018
by Juri Sarbach, CTO – published on Medium.com on 10 November 2018
by Juri Sarbach, CTO – published on Medium.com on 29 November 2018
We engage with data science teams in the exploration, development, and distribution of AI products and services. To empower corporate data science teams, we focus on establishing operational excellence and best practice workflows using the AI Platform on the Google Cloud Platform (GCP) as well as Kubeflow. Together with our strategic partner Panter, we provide services from engineering cloud architectures around AI models to building end-to-end solutions on the GCP.