Why 80% of ML Time Gets Wasted 📉

At #BookingDotCom, 80% of #ML development time went to data preparation.

Why?

  • Duplication: Teams kept rebuilding the same features differently.
  • Training-Serving Skew: Models behaved one way in training, another in inference.

The Fix: Reusable computations for end-to-end consistency.

🗓️ Full talk live on InfoQ Nov 18! Mark your calendar!

#MachineLearning #DataEngineering #MLOps #FeatureEngineering