White Paper: Robotics, Automation & Control

Data Supply Chain for Data Centric AI


The pivot of AI/ML teams toward “Data Centric AI” means more attention on data quality, diversity, and fit for the intended purpose of training and improving model and system performance.

ADAS/AV and IT Infrastructure teams face key challenges when it comes to the large, decentralized datasets critical to delivering the safety, convenience, and efficiency expectations of the autonomous world. Robust yet flexible data pipelines must address key challenges such as:

  1. Processing diverse and complex data types and quickly identify relevant data (usually 1%-10% of ingested data)
  2. Evolving constantly to meet changing data science requirements
  3. Scaling-up and tracking large volumes of data, especially as AI/ML transforms data into code

There is an urgent need for technology that ties data together with data processing infrastructure from edge locations, user data centers, and public clouds. The Akridata Edge Data Platform provides users such a product set to build dynamic and automated pipelines from the edge to core to cloud while leveraging existing infrastructure and software assets, enabling them to track, search, and manage petabytes through exabytes of data.

Measurable results show actionable, visible data can double Data Scientist productivity, lower infrastructure costs by 50% because data is better categorized and assigned to the most appropriate storage tier, and provide access to relevant, cataloged data in hours compared to days or weeks.

Don't have an account? Sign up here.