Data Science: Dataiku Deploys Fast on AWS using CloudStack

Using a no-code accelerator, Dataiku’s ML model and data management platform can be deployed in no time in AWS’s public cloud. Business users and data scientists can access it from their browser and can also connect to AWS-managed cloud services.

A few weeks ago, French publisher Dataiku delivered its code-free CloudStack accelerator to facilitate the deployment of its data management and artificial intelligence platform in the AWS public cloud. Once deployed, it will be accessed by various users, business teams and data scientists from their web browsers to work with the data. Some create reports and dashboards, others are to build machine learning models. This CloudStack accelerator is not charged extra.

There are three steps to deploying the platform on AWS, which Dataiku describes in a post for cloud architects. The first is to create permissions on the existing AWS virtual private cloud topology, for creating Dataiku instances. Next is the deployment, using four out-of-the-box templates with what it takes to start and grow analytics and AI projects and put them into production in the cloud. The third step involves the maintenance, evolution and updating of the Dataiku platform on AWS by the company’s IT team. With CloudStack, deployment time is reduced from a few days or weeks to a few hours, according to its publisher.

(larger image) The four deployment models offer different architectural blueprints: a single-node design environment for building pipelines and data models, or environments for data science team in need of elastic resources (Kubernetes clusters). Credit: My data

Access to AWS services such as Athena, Glue, Recognition …

Once the data management and AI platform is deployed on AWS, business users will prepare their datasets using Dataiku’s no-code visual function or using custom SQL code. Data scientists, for their part, will have access to the latest AutoML version of the platform for their machine learning models. They can rely on AWS EKS (Elastic Kubernetes Service) service. Other features available to manage ML models include data pipeline and model tracking, data rift alerting, A/B testing, and model reformation, Dataiku says. Management functions are also offered to reduce the risks associated with the models. In addition, users of the platform will also be able to connect to a variety of AWS services, including managed services such as Athena and Glue, Comprehend or Recognition.

Men and women of data science

At the end of last year, to raise awareness of the talented data science figures, who have advanced in the discipline and continue to do so, Dataiku launched the History of Data Science site. This is the result of several months of collaborative work. Along with historical figures such as Charles Babbage (1791-1871), who described the principle of the calculating machine, or in the 1950s, the Dartmouth Summer Research Project team that laid the foundations of artificial intelligence, we discovered that engineers and researchers at the heart of innovation in data science today.

Along with portraits, scroll for example profiles Fei-Fei Li, professor of computer science at Stanford University, Yoshua Bengio, one of the pioneers of deep learning, won the 2018 AM Turing Prize along with Frenchman Yann LeCun and Geoffrey Hinton, or Judea Pearl who was awarded the prize in 2011 , “known for having a probabilistic approach to AI and developed by Bayesian networks, as well as the basic algorithms used for prediction in these models”, the narrative in particular History of Data Science. On the site, you can also access the game Beat the Linear Regression Algorithmand we will get a copy of the album Data Science Innovators: From Bayesian to Bayesian Neural Networks.

Leave a Comment