Strengthening Foundational Literacy and Numeracy through Accurate, Scalable Assessments
Case at a Glance
Impact
~16 lakh
students assessed through AI-driven FLN tools
~ 6000
education supervisors mobilised, resulting in 66,000 school visits
39%
reading proficiency vs the 69% reported, triggering a statewide reset of FLN priorities.
About the organisation
Vowels of the People Association (VOPA) builds technology-led, scalable solutions in education and mental health to nurture responsible and empowered citizens. Its flagship initiatives include statewide FLN assessment systems, free digital learning ecosystems, and culturally relevant mental health platforms for underserved communities.
Problem Statement
Official FLN assessment data in Maharashtra indicated reading proficiency levels of 69–74%. However, evidence-backed assessments revealed actual proficiency below 40%, with nearly 60% of early-grade children struggling with basic reading and numeracy. Existing assessment practices relied on superficial or rote-based evaluations, creating false visibility, delaying interventions, and leading to misinformed policy and resource allocation decisions.
Solution
VOPA developed an AI-powered FLN assessment system through the ‘NIPUN Maharashtra’ app to deliver accurate, joyful, and scalable oral reading and numeracy assessments. The solution enables real-time, evidence-led evaluation of student learning using speech recognition, adaptive assessment flows, and automated reporting. Teachers, supervisors, and administrators receive actionable insights through role-specific dashboards, while parents can track assessment evidence and access personalized remedial tools.
Quick Facts
Organisation Name
Vowels of the People Association (VOPA)
Organisation Website
Visit Site
Founding Year
2018
Number of Beneficiaries served
~16 lakh students assessed
Geography Served
Maharashtra
Focus Area
Programmatic Impact Operational Efficiency
Functions Impacted
Monitoring, Evaluation & Learning (MEL) Program Delivery / Beneficiary Services Technology & Data Management
Service Provider
Developed In-house
Contact Email
contact.vopa@gmail.com
SDG Addressed
Full Case Study
Challenge
Inaccurate assessments masked learning gaps, delaying systemic response
Foundational Literacy and Numeracy (FLN) assessments are critical for guiding early learning interventions and policy decisions. However, in Maharashtra, existing assessment practices created a misleading picture of progress:
Inflated performance signals: Official data reported reading proficiency levels of 69–74%, suggesting that foundational learning outcomes were largely on track.
Hidden learning deficits: Evidence-based oral reading assessments revealed that only 39% of students met required reading proficiency, leaving nearly 60% of early-grade children struggling with basic literacy skills.
Superficial assessment methods: Rote-based and low-rigour evaluations failed to measure comprehension, fluency, and application, masking the true extent of learning gaps.
Misaligned system responses: Teachers, administrators, and policymakers planned interventions based on inaccurate data, prematurely shifting focus away from foundational learning.
High opportunity costs: Large-scale assessments consumed millions of person-hours without generating actionable insights, delaying remediation and misdirecting training, budgeting, and monitoring efforts across the education system.
Solution
An AI-driven, evidence-based assessment system for scale and accuracy
VOPA’s AI-powered ‘NIPUN Maharashtra’ app was designed to generate reliable, real-time evidence on FLN outcomes at scale. The solution uses refined open-source Marathi Indic AI models for Automatic Speech Recognition (ASR) to assess oral reading fluency across alphabets, words, sentences, and passages.
Key features of the solution include:
- AI-driven oral reading and numeracy assessments using speech recognition.
- Evidence-led, adaptive assessment flows across alphabets, words, sentences, and passages.
- Real-time audio streaming with noise detection and chunking for low-bandwidth classrooms.
- Automated assessment reports capturing words per minute, accuracy, and error patterns.
- Role-based dashboards for teachers, headmasters, supervisors, and administrators.
- Integration with government systems such as UDISE+, SARAL, and SHALARTH
Rolled out through a government circular, the project was made mandatory for all schools, enabling rapid statewide adoption.
Outcomes & Impact
Revealing reality, reshaping priorities, and enabling action
- Large-scale coverage ~16 lakh students assessed statewide, far exceeding the original sample requirement of 4 lakh students.
- Supervisor mobilisation ~6,000 education supervisors conducted 66,000 school visits, strengthening monitoring and on-ground engagement
- Evidence-led course correction Assessments showed only 39% reading proficiency versus 69% previously reported, triggering a statewide reset of FLN priorities.
- Policy and budgetary impact Findings influenced teacher training plans, departmental budgets, and monitoring frameworks, shaping decisions worth hundreds of crores.
- Granular insights Detailed evidence on reading speed, accuracy, and error patterns enabled targeted academic interventions by grade and geography.
- Adoption indicators 1,87,000+ app downloads and positive feedback from over 1,300 teachers and education officers.
Technology Stack
| Name of the Tool | Where it was used | What it enabled | Category |
|---|---|---|---|
| Marathi Indic ASR Models | Oral reading assessments | Speech-to-text evaluation for FLN | Open-source |
| FastAPI (Python) | Backend framework | Scalable API-driven architecture | Open-source |
| MongoDB | Data storage | Secure storage of assessment data | Open-source |
| Redis | Caching layer | Low-latency performance at scale | Open-source |
| Noise Detection Library | Audio processing | Reliable assessment in noisy classrooms | Open-source |
| WebSockets with Audio Chunking | Live assessments | Real-time, low-bandwidth audio streaming | Open-source |
Key Project Learnings
VOPA’s implementation demonstrates how evidence-led AI can drive large-scale course correction in education systems:
- Evidence Unlocks Action: Replacing inflated, self-reported assessments with AI-backed evidence revealed the true extent of FLN gaps. This visibility enabled the state to reset priorities toward foundational learning and redirect training, monitoring, and budgets at scale.
- Designing for Scale Enables Adoption: Simple, joyful, and low-friction assessment flows allowed teachers and supervisors to adopt the system quickly. This ease of use supported statewide rollout, large supervisor mobilisation, and assessment coverage far beyond initial targets.
- Data Strengthens Accountability: Granular, role-based dashboards linked learning outcomes to action at every level. This improved follow-through, enabled targeted interventions by grade and geography, and strengthened accountability across the education system
Potential for Wider Adaption
| Sector | Adaptability of the Solution |
|---|---|
| State Education Systems | Applicable for large-scale FLN assessments across diverse linguistic and geographic contexts |
| National Education Programs | Relevant for evidence-based monitoring of foundational learning outcomes |
Additional Details
The information provided here is created as a community resource and is not intended as professional advice or a recommendation by ILSS or Koita Foundation. While we strive to ensure the accuracy of the content, we do not take responsibility for any errors or omissions. Users should use their own discretion before making any decisions based on this information. ILSS or Koita Foundation assume no liability for any actions taken based on the information provided.