Personalised Context-Aware Music Recommendations in Real-Time

October 2020 - June 2021

Over the last decade or so, the use of streaming services has skyrocketed, and a key time people listen to this music is when exercising.

However, it's hard to find the ideal songs to listen to when working out.

That's where Osti comes in...


We had 4 key aims for this project, shown in the following image

Objectives - Osti

Techniques Used

Osti uses a hybrid approach of collaborative filtering and content-based recommendations to generate recommendations

Techniques Used - Osti

Microservice Architecture

Osti is split into multiple microservices for fetching data, generating recommendations and displaying data to a client, all written using appropriate tech stacks

Microservice Architecture - Osti


Osti was evaluated and was shown to keep user vitals closer to their target values

Evaluation - Osti

This project was part of my MEng Computing degree at Imperial College London. The full report details all of this in more depth.

Thanks to my supervisor Professor William Knottenbelt for assistance during this project.