AI in Indian education

AI for everyone who teaches, learns, or builds education in India.

TechAnk is a founder-led Indian micro-enterprise. Two AI products already in the market — CheckMate for teachers grading papers, and Qwizee for aspirants preparing for the country’s most competitive exams and the coaching institutes that train them. We do the rote work end-to-end. You make the calls that matter.

A small company with a long horizon.

TechAnk is a Udyam-registered Indian micro-enterprise. We build AI products for the people who run Indian education — teachers grading exams, aspirants preparing for competitive selection tests, and the coaching institutes that train them.

We’re not interested in AI as a layer painted over old workflows. We build products where the AI does the rote work end-to-end and the human stays the final authority on every decision that matters — whether that’s a teacher grading a paper, an aspirant deciding what to study next, or an institute running an exam for hundreds of students. Indian focus today, global aspirations tomorrow.

2025
Founded · Udyam-registered MSME
2
Live AI products · one shipping, one in pre-pilot
asia-south1
All data stays in India

Two products. One mission.

Both products use the same underlying belief: AI should do the rote work end-to-end, and the human stays the final authority. Teachers review marks. Aspirants review their own progress. The AI never decides without you.

CheckMate

Pre-pilot

For teachers. A teacher uploads the question paper once, then uploads each student’s answer sheet from a phone. CheckMate produces per-question marks, feedback, and a parent-shareable PDF report — in 1–2 minutes per sheet instead of 8–12.

Before any student sees the paper, CheckMate also runs a pre-flight quality review of the question paper itself. In a recent pre-pilot session it caught seven mistakes in a single Class 11 Chemistry paper from a ten-school chain — variable mismatches, reversed activity-series facts, marks totals that didn’t add up. All flagged before the paper reached students.

  • Handles Punjabi, Hindi, English, mixed-script answers
  • Half-marks, OR-groups, and sub-parts modelled correctly
  • Pre-flight paper quality review catches teacher mistakes before students do
  • Teacher always reviews and can override any grade
  • Audit trail on every override; full transparency for parents
  • Built on Google Cloud · all data processed in Mumbai
Visit aicheckmate.in
CheckMate — paper checking that gives teachers their evenings back.
Qwizee — AI-powered practice for India’s most competitive exams.

Qwizee

Live on Play Store

For aspirants preparing for India’s most competitive exams — and the coaching institutes that train them. Live today on Google Play with Punjab Government tracks and a live-exam runner that coaching institutes already use to conduct their own practice tests. v2 expands to Bank, UPSC, and TET, with AI essay evaluation built on the same grading core that powers CheckMate.

  • Live today: Punjab Government tracks on Google Play
  • Live today: Coaching institutes run their own live practice tests on Qwizee’s infrastructure — no separate exam platform needed
  • v2: Expanding to Bank (IBPS, SBI), UPSC, and TET preparation
  • v2: AI essay evaluation for UPSC aspirants — instant feedback, same grading core as CheckMate
  • v2: Daily news, learning section, and AI-generated practice tests calibrated to your weak areas
Download on Google Play

Not a Gemini wrapper. A continuous-learning system.

The LLM under the hood is a commodity. The differentiator is the labeled corpus that accumulates from every teacher correction across every pilot school — and the pipeline that turns that corpus into quarterly improvements to a Gemini 3.5 Flash model fine-tuned specifically for Indian board exam evaluation.

We do not believe in fine-tuning on synthetic or public-only corpora; the entire value of this architecture is that it captures real teacher pedagogical intent. Every pilot evaluation a teacher reviews adds labeled examples. Six months of pilot data produces a more valuable corpus than any competitor could buy. It compounds.

3-step
LLM perceives. Code structures. Teacher confirms.
~$72
Cost per quarterly retrain on pilot-scale corpus
10 → 50 → 100
A/B traffic ramp for every new tuned model

Built on Google Cloud — Vertex AI for generative inference and supervised tuning, Firestore as the labeled-corpus store, Cloud Functions for orchestration, all in asia-south1. Customer data stays in India by default; never used to train upstream Google models.

One founder, ten years of engineering, one focus.

A

Ankur Kaushal

Founder · TechAnk

Ten years as a backend engineer across Java, distributed systems, and enterprise SaaS. TechAnk is the long-term home for the AI products I’m shipping for Indian education — Qwizee live on Google Play, CheckMate in pre-pilot with the first school launch scheduled for 2 June 2026, more in the pipeline.

Building deep AI products on Google’s stack — Vertex AI, Cloud Functions, Firestore, all in asia-south1 for in-country data residency. Currently applying CheckMate’s continuous-learning architecture (every teacher correction trains the next quarterly model) to Indian board exam evaluation, with global expansion to follow.

For schools, partners, and curious people.

If you run a school and want to try CheckMate in your pilot batch, or you’re an investor, mentor, or fellow founder building in this space — write to us.

ankur@techank.com

TechAnk · India · Replies usually within 24 hours.