Platform Architecture

One Platform for the Entire AI Lifecycle

  Ingestion & Preparation

  • Scalable batch & streaming ingestion
  • Visual data flow editing and provenance
  • Monitor feed health in real time
  • Search metadata, lineage & stats

Discovery & Visualization

  • Full-text, faceted & geospatial search
  • Real-time, drag & drop dashboards
  • Interactive time series analysis
  • Dataset ingestion & governance
  • Build, share & annotate dashboards

 Modeling & Experimentation

  • Machine Learning & Data Mining
  • Train & debug on CPU’s & GPU’s
  • Natural Language Understanding
  • Performant machine learning at scale

The Data EngineerThe Data AnalystThe Data ScientistThe DataOps EngineerThe App Developer

 Model Productization

  • Auto-test & deploy machine learning
  • Reusable & versioned model store
  • PyTorch, TensorFlow, R, Spark, Scikit
  • Security, logging & rate limiting

 Operations

  • Portal & Single Sign on
  • Container Orchestration
  • Live Monitoring
  • Authentication, Authorization, Audit


 Design end-to-end data pipelines with close to 300 processors.

A view of one processor configurations that is used to List objects within a specific path in an S3 bucket.


Visualization options that can be created in SQL Analytics.

Explore data via free text search, then shape your data via charts, tables, maps, and dashboards.

An example of using Spark NLP’s Named Entity Recognition (NER) OCR algorithm and aggregating the results across multiple TCGA oncology reports.

An interactive environment for dealing with relational databases.


Notebook environment where data scientists can work with the data and publish Machine Learning models.

R based notebooks.


A common portal for accessing all applications.


Alerting channels available for system admins of the platform.

Admins can manage the state of the system interactively from the same portal.

Admins can measure metrics flowing in from the cluster manager and set alerts.

Manage users, roles, and group permissions as well as authentication flows for applications.

Users can enable two factor authentication.