Data Sources
Amazon Infrastructure
Junction TV Platform


MI-3 Platform

MI-3 platform is built to scale massively in supporting 10B+ events per day. The underlying data pipeline provides high bandwidth data ingest, in-stream Extraction, Transformation and Load (ETL), security, and data reliability. Platform is deployed on Public (AWS) and on-premise Cloud. Via optimizing underlying compute resources, the platform reduces infrastructure cost while achieving scale.

Data Pipeline Architecture

Our analytics platform is based on modern Big Data Lambda Architecture, which combines both real-time and batch processing systems under one highly scalable query platform. It enables processing of data scaling from 10M+ devices and billions of device events/sec. Our data pipeline is both programmable and extensible.

End-to-End Automated and Orchestrated

Managing the end-to-end data pipeline and processing platform makes operation streamlined and reliable. Our cloud management system automates this process via resource monitoring, automating resource scaling, stitching the service constructs and managing the resource access control/security. The orchestration system supporting multi-tenant deployment can build and deploy the end-to-end platform in 1 hour resulting in fast turn-over and ability to on-demand create analytics processing system.

Unifying Analytics across all data sources

Today, media analytics services are fragmented. There are too many data silos, lack of correlation and contexts. One service consumes data only for recommendation and discovery, while another for analyzing subscriber engagement. Such fragmentation leads to lack of insights, lack of actionable results, higher information management overhead. Our approach is to unify the data from multiple sources into a single data pipeline and providing the ability to build any analytics applications.

Data Models

Data models are fundamental building block of MI-3 platform. Data models define the structure for building and using the data-driven intelligence. The model is rich and flexible to build on and evolve multiple analytics applications. Data Models are abstracted through APIs as well as query frameworks.

Advanced Algorithmic Engines

MI-3 platform analytics engine is a composition of multistate advanced algorithms. These engines are designed to target the TV and Media space to extract intelligence and deep insights. We have developed specific correlation algorithms, classification algorithms and deep learning algorithms that provides an array of services such as content recommendation and churn prediction.

Open and Programmable Reporting

Unlike most analytics service providers who are limited to dashboard full of graphs, we provide open APIs for building your own dashboard and further programmability on reports. Furthermore, we provide data warehousing on top of Amazon Redshift allowing you to connect to various data integration and business integration tools and services.

Self-service and Streamlined Integration

Analytics services in TV/ Media space is continuously evolving with the need for new insights specific to type of services. We want to enable such evolution by avoiding lock-ins. MI-3 Platform allows operators to leverage our platform to build their own data processing system. We further provide various adaptor and device SDKs to connect data sources.








It can be as simple as


Provision an end-to-end data processing system in 1 hour


Integrate via adapter, Rest APIs or device SDKs


Begin consuming analytics services


Evolve: add more data, contexts, create more services