Recommendation System: Solving cold start problem using Graphical Neural Networks

I'm looking for an experienced developer who can help me build a recommendation system to solve the cold start problem. The solution should involve collaborative filtering, and the system should not require real-time recommendations. The data necessary for this model is tabular categorical data. The developer should have experience in creating recommendation systems, as well as the technology and workflows required to build the system. The successful candidate will be able to create a solution that meets all of my goals quickly.
The dataset is simple and contains just 4 columns, namely, userID, productID, ratings and timestamp.