Exercice of the Cohort 2 of Building a Real-Time ML System Together by Pau Labarta Bajo

Summary

This is a real world practice to improve MLOPS and ML engigneering skills. The objective it to build a ML system for the crypto price prediction problem. Even if i've writen 70% of the code in this repository (me and my LLM friend) to ensure I learn the topics, I must say that the original repostitory has been a guide in the 99% of the steps, it is here https://github.com/Real-World-ML/real-time-ml-system-cohort-2.

Tech stack

• docker
• redpanda(Kafka)
• quixtreams
• hopsworks
• comet

the journey

• Build independent Microservices
• Deal with streaming data and kafka topics
• RESTAPI and WebsocketAPI conections
• Pushing and retrieving topics to and from the feature store
• dockerize real-time feature pipeline
• dockerize backfill pipeline
• build a training pipeline
• feature engineering
• improve the model with CV hiperparamenter tuning ...