Data Scientist / ML Engineer

Conducts correlation analysis to identify multicollinearity; assesses feature importance using techniques like recursive feature elimination or tree-based methods; determines which features to keep based on model impact; considers dimensionality reduction techniques; plans to retrain the model with selected features.

Identify the source of corruption and affected records; assess the impact on model performance; either remove problematic data or substitute with reliable data; communicate findings to the team; prepare for re-training if necessary.

Indicates moderate variance in model performance across folds; considers methods to reduce variance such as increasing training data or simplifying the model; evaluates feature selection techniques; may experiment with different cross-validation strategies.

Evaluates the current bottlenecks in data processing and model inference; considers parallel processing or batch processing techniques; assesses infrastructure for scalability options like cloud solutions; prioritises optimising code and algorithms before scaling hardware.

Recognises that the MAE indicates average error magnitude, while the MSE penalises larger errors more heavily; considers whether the distribution of errors is skewed; discusses potential outliers affecting MSE more than MAE; suggests investigating residual plots to identify patterns.

Names the specific unexpected results and their impact on business; discusses the process followed to investigate the root cause; details any changes made to the model or data; mentions communication with stakeholders and lessons learned; avoids blaming external factors.