← Dow Interview Insights

Dow·Software Engineer·Technical Phone Screen·Senior

Senior
Apr 2026

Summary

Research-focused interview at Dow for a Research Engineer role, centered almost entirely on SHAP analysis and its behavior on nonlinear physical descriptors. Pretty deep technically, the kind of conversation where you feel like you're being tested on whether you've actually hit the limits of a tool or just read the docs.

Questions Asked (3)

Q1

Does SHAP's additive consistency property still hold in a meaningful way when applied to highly nonlinear force-field descriptors like SOAP?

Technical Trade-offsRoot Cause Analysis
Author's notes

This is where I felt the ground shift a little.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Clarify that SHAP's additive consistency is a mathematical property of the model's output decomposition, not of the input features, so it holds for any function including SOAP-based models. Then discuss the practical implications: while the property holds, the interpretation of SHAP values for highly nonlinear descriptors may be less intuitive due to feature interactions and the high-dimensional nature of SOAP. Finally, relate this to the trade-off between model interpretability and predictive accuracy in materials science.

Pro tip: Acknowledge that SHAP values for SOAP descriptors can be dominated by a few components, and suggest using aggregated SHAP values over SOAP's local environment basis to gain chemical insight, showing you understand both the theory and the application.

1. Define SHAP's additive consistency

Explain that SHAP values sum to the difference between the model's prediction and the expected value, and that consistency ensures that if a feature's contribution increases, its SHAP value does not decrease.

2. Clarify the scope of the property

Emphasize that additive consistency is a property of the model's output decomposition, independent of the input feature type or model nonlinearity.

3. Address SOAP's characteristics

Describe SOAP as a high-dimensional, rotationally invariant descriptor that encodes local atomic environments, often used with nonlinear models like neural networks or Gaussian processes.

4. Discuss practical interpretability

Explain that while the property holds, the SHAP values for individual SOAP components may be difficult to interpret due to correlations and interactions, but aggregation or dimensionality reduction can help.

5. Relate to trade-offs and root cause analysis

Connect to the role by highlighting the trade-off between using expressive descriptors for accuracy and maintaining interpretability, and how SHAP can aid root cause analysis in materials design.

Key Points to Mention

  • SHAP's additive consistency is a mathematical guarantee for any model function.
  • SOAP descriptors are high-dimensional and encode complex local environments.
  • Nonlinearity does not break the property but can affect interpretability.
  • Feature interactions in SOAP can lead to non-additive effects in the model, but SHAP still provides additive attributions.
  • Aggregating SHAP values over SOAP components or using kernel SHAP with a background dataset can improve interpretability.
  • In industrial contexts like Dow, interpretability is crucial for trust and root cause analysis, so combining SHAP with domain knowledge is key.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

Can applying SHAP to nonlinear physical descriptors introduce overfitting or feature-redundancy artifacts, and why?

Technical Trade-offsRoot Cause Analysis
Author's notes

Blanked for a second on the overfitting angle specifically.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Acknowledge that SHAP is a model-agnostic interpretation tool and its outputs are only as reliable as the underlying model and data. Explain that nonlinear physical descriptors can indeed lead to overfitting or feature-redundancy artifacts in SHAP values, especially when descriptors are correlated or the model overfits. Then discuss how to detect and mitigate these issues through validation and careful feature engineering.

Pro tip: Emphasize that SHAP explanations should be validated against domain knowledge and held-out data; if a SHAP value contradicts physical intuition, it's a red flag for artifacts. Also, mention that using SHAP with regularized models or on a separate interpretation set can reduce overfitting.

1. Clarify SHAP's role

State that SHAP explains model predictions, not the underlying physical system. Artifacts arise from the model or data, not SHAP itself.

2. Identify sources of artifacts

Discuss how nonlinear descriptors can cause overfitting if the model is too complex, and feature redundancy if descriptors are correlated, leading to misleading SHAP values.

3. Explain mechanisms

Describe how overfitting leads to unstable SHAP values that don't generalize, and how redundancy splits importance among correlated features, making interpretation difficult.

4. Propose detection and mitigation

Suggest using cross-validation, regularization, feature clustering, or grouping correlated features before computing SHAP. Validate SHAP with domain knowledge.

5. Conclude with best practices

Summarize that while SHAP is powerful, it requires careful model selection and data preprocessing to avoid artifacts, especially with nonlinear physical descriptors.

Key Points to Mention

  • SHAP values are model-dependent and inherit model biases.
  • Nonlinear models (e.g., tree-based, neural networks) can overfit to noise in physical descriptors, leading to unstable SHAP values.
  • Correlated features cause SHAP to distribute importance arbitrarily, creating redundancy artifacts.
  • Use of regularization, cross-validation, and feature grouping can mitigate these issues.
  • Validate SHAP explanations with domain expertise and held-out data.
  • Consider alternative interpretation methods or simpler models when artifacts are suspected.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

What are the core limitations of SHAP as a method, and how would you go beyond just using the library to improve attribution quality for nonlinear physical descriptors?

Technical Trade-offsSystem DesignAlgorithms & Data Structures
Author's notes

The 'beyond library-level usage' framing is what made this interesting.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging SHAP's theoretical guarantees and practical utility, then systematically outline its core limitations (computational cost, approximation errors, correlation handling, and linearity assumptions). Transition to concrete strategies for improving attribution quality in nonlinear physical systems, such as using domain-specific kernels, hybridizing with physics-based models, or developing custom attribution methods that respect known invariances.

Pro tip: Emphasize that in industrial settings like Dow, attribution must be actionable and physically plausible; propose validating SHAP explanations against domain knowledge or controlled experiments to ensure they drive real process improvements.

1. Acknowledge SHAP's strengths and context

Briefly state SHAP's value (e.g., Shapley values' theoretical foundation, model-agnostic nature) to show balanced understanding before critiquing.

2. Enumerate core limitations

Discuss computational complexity (exponential for exact, sampling variance for approximations), challenges with correlated features, and assumptions of feature independence/additivity that often fail in physical systems.

3. Highlight issues specific to nonlinear physical descriptors

Explain how nonlinear interactions and physical constraints (e.g., conservation laws, monotonicity) can lead to misleading attributions when using standard SHAP.

4. Propose enhancements beyond the library

Suggest methods like custom kernel SHAP with domain-informed weights, integrating physics-based surrogates, or using causal inference to handle correlations.

5. Validate and iterate

Describe how to validate improved attributions via domain expert review, synthetic data with known ground truth, or stability tests across perturbations.

Key Points to Mention

  • Computational complexity: exact SHAP is exponential; approximations introduce variance.
  • Correlation vs. causation: SHAP attributes to correlated features can be misleading.
  • Additivity assumption: SHAP assumes additive feature contributions, which may not hold for nonlinear physical interactions.
  • KernelSHAP and DeepSHAP: their limitations in high-dimensional or highly nonlinear settings.
  • Domain-specific adaptations: incorporating physical constraints (e.g., monotonicity, conservation) into attribution.
  • Validation strategies: comparing attributions with known physics or using ablation studies.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.