Open sourceFree · no sign-up

Good recommendations from day zero, even for brand-new users.

Adaxon Recommender. Hybrid recommendation engine that solves cold start.

View the code ↗
adaxon-recommender
AR
Why it exists

01PROBLEM

Recommenders fail new users with no history, and show them generic popular items.

02SOLUTION

Weighted matrix factorisation (ALS) for known users, plus country and platform cold-start segments. Containerised.

03VALUE

Higher early engagement, and a production-grade reference design.

Screenshots

The story

How to use it

  1. Run the containerised service.
  2. Feed it interaction data such as clicks and installs.
  3. Request recommendations for known and brand-new users.

How it's built

A hybrid recommender: weighted matrix factorisation (ALS) for users with history, plus country and platform segments to handle new users. Packaged with Docker.

Why it matters for businesses

New users who get relevant recommendations engage sooner, which improves early conversion and retention.

In real life

An app store shows a first-time visitor in India on Android a relevant set of apps on day zero, instead of a generic top-ten list.

Under the hood

How it's built.

// adaxon-recommender · Data & ML
export const adaxonRecommender = {
stack: ["Python", "ALS", "Docker"],
audience: ["ML engineers", "Growth teams"],
status: "open-source", price: 0
}
Shape what's next

Feature requests.

No requests for Adaxon Recommender yet. Be the first!

Chat with Bala