EdTech

An adaptive tutoring engine for 240k students

A personalization engine that lifted average student outcomes 18% on standardized assessments — built in 12 weeks for a Series-B EdTech.

Avg outcome lift
+18%
Students live
240k
Teacher approval
94%
MVP to GA
12 wks
The challenge

Existing personalization engines optimized for engagement metrics, not learning outcomes. Cohort needed an engine teachers could trust to align with curriculum standards.

Our approach

A Bayesian knowledge-tracing model layered with an LLM-driven explanation generator. Per-curriculum constraints enforced by policy as code.

The outcome

Pilot districts saw 18% improvement on year-end assessments. The system became Cohort's marquee Series-B differentiator.

Their team understood pedagogy as well as our pedagogy team. That's not what I expected from an AI engineering firm.

James Okafor · VP Product, Cohort.io
Project details
Client
Cohort.io
Industry
EdTech
Duration
12 weeks
Region
London, UK
Tech stack
PythonPyTorchPostgresGPT-4oModalVercel
Services delivered
AI Product DevelopmentGenerative AIEmbedded AI Team

Similar challenge?

Let's scope your project

Book a call and tell us what you're trying to ship. A senior engineer will respond within one business day.