Robin (Rob) Kras

Data Engineer — Databricks, Azure, and scalable data infrastructure

Leiden, Netherlands · robkraseu@gmail.com · LinkedIn · GitHub · Kaggle

Summary

Data Engineer working across Databricks, Azure, and the data pipeline from raw data to production. Focus on MLOps, infrastructure as code, and cloud-based data engineering.

Experience

Data Engineer
Rabobank
  • Build and maintain scalable data pipelines and ML infrastructure in production environments.
  • Specialize in MLOps practices, containerization, CI/CD automation, and cloud-based data engineering.
  • Own application maintenance, production deployment via infrastructure as code, automated testing, and monitoring systems.
Research: Multimodal Sound Symbolism
Leiden University
  • Investigated cross-modal sound symbolism in vision-language models.
  • Combined computational linguistics, computer vision, and cognitive science. Grade: 8/10.
  • View on GitHub

Education

MSc Computer Science — Data Science & AI
Leiden University
  • Thesis: Cross-Modal Sound Symbolism in Vision-Language Models
  • Grade: 8/10
BSc Computer Science — Minor: Data Science
Vrije Universiteit Amsterdam

Technical Skills

CategorySkills
Data & AnalyticsDatabricks, Apache Spark, SQL, Python, Pandas, Azure Data Factory, Azure Synapse
ML & AIMachine Learning, MLflow, XGBoost, LightGBM, CatBoost, PyTorch, SHAP
Cloud & DevOpsAzure, AWS, GCP, Terraform, Docker, Kubernetes, CI/CD, Infrastructure as Code
LanguagesPython, SQL, Scala, Bash, C/C++

Notable Projects

Rainfall Prediction
  • Feature engineering, K-Folds cross-validation, and ensemble methods.
  • Simpler algorithms (KNN) outperformed complex ensembles once properly optimized.
House Prices Prediction
  • Regression techniques with domain knowledge and SHAP for feature importance.
Loan Payback Prediction
  • Binary classification for financial risk assessment.
  • Ensemble of XGBoost, LightGBM, and CatBoost with SHAP interpretability.
Music BPM Prediction
  • Predicting song tempo from audio features, combining signal processing with ML.
Road Accident Risk
  • Ensemble methods with temporal and weather interaction features; containerized with Docker.
Bank Marketing
  • Customer response prediction with YAML-driven configuration for rapid prototyping.

Additional Projects

ProjectKey Techniques
Podcast Listening TimeTime Series, Stacking, Temporal Features
Optimal FertilizerMulti-label, MAP@3, Domain Knowledge
Titanic SurvivalXGBoost, Feature Engineering
Credit Card FraudImbalanced Data, SMOTE, K-Folds
Personality TypeSMOTE, Bayesian Optimization

Open Source

ML Utilities Library
  • Reusable machine learning components and helper functions for data science workflows.
  • View on GitHub