Model Serving

Real-time inference from the model is not necessary. Once the user submits a request, the model does not need to consider the new input in real-time. Hence, the estimation can be pre-computed on a nightly basis and stored in cache.