Skip to content
ASIP Mentor guide
AI, data and signal processingB06Partner brief2 of 12 in AI

Finding a Child's Real Learning Level

Project SparkL

A child enrolled in Grade 6 may read at Grade 2. Find the real level, at scale.

Skills
  • Speech and audio processing
  • Adaptive assessment logic
  • Offline-first design
  • Indic-script interfaces
Domains
Education

Who you work with#

Project SparkL is a Bengaluru trust, registered in 2022, that teaches foundational reading and arithmetic to children from vulnerable backgrounds, including children living in shelter homes. It runs after-school evening sessions at seven centres for about 130 to 140 children, with two partner organisations.

The situation#

The children SparkL works with are enrolled in a school grade that does not match what they can actually do. A child may be registered in Grade 6 and be unable to read a Grade 2 sentence.

SparkL already handles this properly. It assesses each child in Kannada, Maths and English across grade levels, works out a “skill grade”, and builds a learning plan from that rather than from the school grade. This is the core of their method and it works.

What is going wrong#

The method is manual, so it stops at seven centres. Assessing each child, working out the skill grade, planning the teaching and tracking progress all depend on a small number of people doing it by hand.

SparkL wants to reach more children. Right now growth means hiring proportionally, which is the definition of a model that will not scale.

Why it is not fixed yet#

Off-the-shelf learning apps assume the child’s grade is their level, which is precisely the assumption that fails here. Adaptive learning products assume a smartphone per child, reliable internet and a literate parent nearby. None of those hold in a shelter home in the evening.

There is also a harder problem. Assessing a child in three languages, including Kannada, without a trained adult listening to them read, is genuinely difficult.

Skills and Expertise#

  • Branch — CS, IT.
  • Technical skills — Offline-first application design, speech or audio processing for reading assessment, adaptive logic, multilingual and Indic-script interfaces, data privacy.
  • Other skills — Working with educators, assessment design, writing for children, understanding a method before automating it.