Natural Language Processing for Exchange Data International

  • Company Name
    Exchange Data International
  • Industry
    Financial Services
  • Location
    London, England

Natural Language Processing for EDI


Add top notch AI based insights from the deluge of unstructured data such as news articles, video, audio, and images to Novus' Portfolio Intelligence Platform. InfinStor's breakthrough Machine Learning platform makes it possible to manage immense amounts of data and cost effectively utilize multiple clouds to create models with accuracy never seen before. As an example, these next generation AI Models can perform Natural Language Processing on news articles with remarkable accuracy.


A NLP pipeline that runs in the InfinStor multicloud ML platform. Directly use pretrained models such as BERT, or fine tune these models with custom data. Deploy model for batch inference of real time inference.

InfinStor ICE Unique Features:

InfinStor Compute Engine (ICE) has the capability to schedule and run machine learning tasks in different regions of possibly different clouds. InfinStor can do this for both cost savings and for enhanced capability. For example, in this solution, our customer was performing data collection activities in India and the Mumbai region was a natural storage location for the collected raw data.

However, the Mumbai region of AWS does not have the latest Nvidia GPU enabled instances. InfinStor ICE solves this problem elegantly by pre-processing the data in the ap-south-1 (Mumbai) region and subsequently running the machine learning inference task in us-east-1 region Nvidia equipped instances.

InfinStor ICE automagically runs preprocessing in Mumbai region CPU instances and Inferencing in us-east-1 region Nvidia instance

InfinStor's InfinSlice helps to manage large amounts of data by providing the capability to micro batch and run inference on only newly ingested data. InfinSnap is the industry's only no-copy solution for data provenance.

Customer Benefits

In the case of EDI, the original proposal was to install two GPU enabled instances in a co-location site at the cost of about $50K upfront, with additional monthly co-location charges. Instead, they chose to use the InfinStor multicloud ML platform, and their cloud bill is approximately $60 per month!

Download White Paper