EDUCATIONAL INTELLIGENCE PLATFORM

Welcome to the future of education!

At Embibe, we have just one mission – to truly personalise education. Because every child deserves it. This has led us to embark on this noblest of journeys to deliver life and learning outcomes for every student! Rooted in consumer behavior, we are leveraging AI to deliver personalised achievement journeys for every student.

Embibe has traversed a long journey from a data-centric product to an AI platform. On this journey, we have realised that the most powerful teams are: 1. Vision Led in understanding student context and obsessed with success; 2. Self-Organising in defining their own agenda; 3. Intellectually Fierce and Globally Conscious in their choices, and 4. Consistently Excellent in their execution.

After exploring a deeply functional organisational structure in engineering, we are now evolving towards a problem-solving team structure that manifests at the platform and backend level as an agile team supporting a unified front-end and augmented by a strong Architect + Principal Engineer + Advisory Group for technical mentoring. This document outlines the problem statement and other aspects of the Educational Intelligence Platform.

THE PROBLEM STATEMENT

We want to build a highly scalable and low latency Educational Intelligence Platform, which can give actionable insights to students, teachers and parents to improve learning outcomes. Typically students’ success is measured by how well they score in exams. At Embibe, we have identified a large number of lateral attributes that can be categorised in two buckets, academic attributes and behavioural attributes. Each student has a story and the content context of that story lends every data query a different meaning. We want to be the education intelligence platform powering real time, conversational AI access to the state of global education. This starts with parents, our school partners and scales to governments.

The platform will efficiently retrieve data processed with models designed around Embibe’s proprietary IP.

We envision a real-time and efficient Educational Intelligence Platform handling TBs of data with the lowest possible cost of ownership to process academic and behavioural attributes and events on a daily basis. This platform will be built with the latest world class technologies. We also endeavour to give back to the community by open sourcing our technology for the education community outside of Embibe.

THE INSPIRATION

We are inspired by:

  • Real Time Analytics of Druid
  • Big Data Processing Pipeline at Uber
  • Airflow by AirBnB

THE OBJECTIVES

  • To c apture and process billions of academic and behavioral attribute events on a daily basis
  • To build a scalable ETL pipeline to cleanse the data to be consumed in a coherent way
  • To store and retrieve an aggregated summary of the academic and behavioral attributes
  • To designing, build and operating a data lake for all the critical data at Embibe
  • To build a scalable event processing pipeline for real-time analyses
  • To build a scalable ETL processing pipeline for the coherent analysis
  • To implement the right level of access management for different roles
  • To ensure Personal Information Protection (PIP) for students, parents, teachers, school, etc.

PRODUCT MANIFESTATION OF YOUR EFFORT

  • School App Intelligence
  • Track
  • Students App Intelligence
  • Timeline
  • My Progress
  • Concept Mastery
  • Spontaneous Success Stories
  • Others
  • Internal Analytics Dashboards
  • U sed Metrics for Company
  • Content Interaction Metrics
  • Others

BUSINESS MANIFESTATION OF YOUR EFFORT

  • Higher NPS
  • Higher Word of Mouth
  • Lower Churn
  • Clear Priorities

METRICS YOU WILL OWN AND LIVE BY

METRIC NAME UNIT FREQUENCY
PROCESSING COST/TB OF DATA USD Monthly
99 PERCENTILE LATENCY OF THE SERVICE LAYER MSECS Daily
ERROR RATE (500’s) AT THE SERVICE LAYERS Percentage Daily
SERVICE UPTIME Percentile Monthly
AVG TIME TO FIX THE P0/P1 PRODUCTION ISSUE Hours Monthly
AVG TIME TO FIRST RESPONSE ON THE P0 PRODUCTION ISSUES Minutes Monthly
AVG TIME TO CLOSE ON THE RCA TASKS FOR P0/P1 PRODUCTION ISSUES Days Quarterly

L2 PROBLEMS OWNED

We believe in building an organisation at the intersection of domain modelling and problem intuition. While the L1 teams give us the flexibility to have a multi-faceted view of the problem and cluster similar problems together, the L2 structure ensures independent and focused problem-solving. The following L2 teams have been suggested for the L1 problem stated above:

  • Data Persistence, Processing and Access Layer : To create a unified data storage and querying platform to support simple reads to complicated aggregations
  • Dashboards and Data Requests – Service Layer : To support easy-to-use plug-and-play data visualisations for content, users, devices, application and infrastructure logs, clickstream engagement, company metrics

L1 SKILLS REQUIRED

  • Product Thinking – Ability to Visualise and Understand Customer Journeys
  • Ability to Draw Insights from Trends and Metrics
  • Experience With the Latest Big Data Technologies Like Hadoop/Spark and Their Ecosystem
  • Experience in Designing Data Lakes and Data Warehousing
  • Working Knowledge of Stream Processing Systems Like Kafka and Lambda Architecture
  • Working Experience in Java/Python/Scala
  • Experience in Building Analytics Platforms That Support Stateless, Complex Aggregations
  • Experience in Implementing OLAP Servers for Large Scale Enterprise
  • Experience in Building Scalable Data Pipelines Using Kafka
  • Previous Experience in Designing and Implementing Event Driven Architecture

IP DEVELOPED SO FAR

  1. Success Stories – Behaviour on Learning Outcomes
  2. Concept Mastery – Knowledge Tracing
  3. EDM Research Paper on Concept Mastery

To Join the Tribe, send us an email on [email protected]