First, name it: this is a classic overfitting curve - the model starts memorizing training noise. Before changing anything, check the validation split for leakage and distribution mismatch.
Data level: Add more data, use data augmentation, check label noise.
Model level: Reduce capacity, add regularization (L2/Dropout), simplify features.
Training level: Early stopping (watch validation loss with patience), lower learning rate, larger batch for smoother gradients.
Suggested order: data, then model, then training. Presenting in this order shows good engineering habits.
Get asked follow-ups live, then receive a scored report — like a real MLE interview loop.
Start AI mock interview