AI EngineeringEdTech

AI-powered personalization that lifted course completion by 47%

Shipped a RAG-based recommendation pipeline into LearnPath’s LMS for 100k+ students, with sub-200ms responses.

LearnPath AI

47%
More completions
38%
Better retention
< 200ms
AI latency
100k+
Students served

The challenge

LearnPath needed to personalize learning paths for 100,000+ students using AI, but lacked the ML infrastructure, expertise, and integration know-how to implement it.

Our approach

  • Built a retrieval-augmented generation pipeline with vector embeddings over their course content.
  • Integrated an LLM for learning recommendations and tuned prompts against real student outcomes.
  • Deployed low-latency inference and wired it into the existing LMS without a rebuild.

What we shipped

A RAG recommendation pipeline with pgvector embeddings and an LLM, integrated into LearnPath’s LMS with sub-200ms inference.

Stack

PythonFastAPILLM APIsPostgreSQLpgvectorAWS

We had the idea and the data but no path to ship it. They built the whole pipeline and plugged it into our LMS — completion rates moved within a month.

Sara IyerProduct Lead, LearnPath AI

Have a similar challenge?

Tell us what you're building. You'll hear from a senior engineer within one business day.

Start a project →