Slashing Grading Turnaround by 95% at TestnTrack with Dual AI Engines
Executive Snapshot
Educational Assessment Revolution
“AI-grading let our teachers focus on teaching, not tallying bubbles.”
Co-Founder, TestnTrack
Challenge & Objectives
Challenge Narrative
TestnTrack relied on hundreds of part-time teachers who manually tallied OMR bubbles, read every handwritten answer, and entered marks. A single mid-term (10,000 papers) still took 3 days to publish results, chewing through overtime budgets and frustrating parents who wanted instant feedback. Scaling to new districts meant hiring more graders, not better tech.
Objectives (Baseline → Target)
Solution Overview
Hybrid AI Assessment Engine
Service Pillars Engaged
Dual AI Engines
1. OCR/OMR Engine
Vision model detects sheet layout, classifies filled bubbles, returns JSON map with 99% accuracy—no dedicated scanners required.
2. Subjective-Answer Engine
Vision-transformer + LLM layer extracts handwriting, decomposes model answer into key-points, awards marks for matches, and spot-scores extra insights.
Assessment Dashboard
Architecture & Approach
Implementation Strategy
1Technical Build
2Discovery & SOP Design
3Deployment & Adoption
Model Improvement Flow
Outcomes & Business Impact
| Metric | Pre-AI | Post-AI | Δ |
|---|---|---|---|
| Result turnaround | 72 h | 4 h | ▲ 95% faster |
| Grading cost / sheet | ₹ 20 | ₹ 3 | ▼ 85% |
| Teacher hrs / 1,000 papers | 100 h | 10 h | ▼ 90% |
| Monthly OPEX | ₹ 12 L | ₹ 1.8 L | ▼ ₹ 10.2 L |
| Accuracy vs. audit | 93% | 98.50% | ▲ 5.5 pts |
| 3-yr projected ROI | — | 640% | — |
Ready to Transform Your Business with AI?
Book a free AI Readiness Audit and get a bespoke roadmap in 7 days.