Skip to content

Expertise

What I bring

Five interconnected areas of expertise, from raw data to production systems. Each one backed by real-world experience at scale.

Selected work

Booking.com · Analytics Engineer · Jun 2025 – Present

Booking.com for Business data platform

Built the Booking.com for Business data infrastructure from scratch.

  • Dimensional models and reporting layers in dbt
  • Pipeline orchestration with Airflow
  • Dashboards for executive and operational reporting
  • ML models deployed to production
  • Integrations with downstream platforms
dbt Airflow Dimensional modeling ML in production
bol.com · Senior Data Analyst · Aug 2023 – May 2025

Brand Insights for strategic advertisers

Led data & insights for the platform's largest strategic advertisers.

  • Custom analyses and executive-level reporting for clients like Procter & Gamble, Unilever and LEGO
  • Marketing measurement frameworks
  • Scaled the Brand Insights program
Executive reporting Marketing measurement Custom analyses
Sources APIs · DBs Staging Rename · cast Transform Logic layer Marts Dim. models Reporting Dashboards dbt Airflow Snowflake
01

Data Engineering

Building data infrastructure end-to-end: ingestion, transformation layers, dimensional models, and orchestrated pipelines across cloud platforms. From raw events to production-ready data assets.

dbt Airflow Snowflake BigQuery SQL Docker

How it works

1

Source integration

Connect to APIs, databases, event streams, and third-party platforms. Centralize source definitions for full lineage.

2

Layered transformation

Staging for cleaned 1:1 source models, intermediate for business logic, marts for dimensional models, and reporting for consumption.

3

Testing and documentation

Automated data quality checks at every layer. Schema tests, freshness monitoring, and auto-generated documentation.

4

Orchestration

Scheduled and event-driven pipelines with dependency management, alerting, and retry logic.

How this helps

A single source of truth for all reporting and analytics, eliminating conflicting numbers across teams.

Self-service analytics: well-modeled data means stakeholders can answer their own questions.

Reliable, tested pipelines that run daily without manual intervention.

Full lineage and documentation so anyone can understand what the data means and where it comes from.

Revenue Users Retention Quarterly Revenue Q1 Q2 Q3 Q4 Trend
02

Analytics & BI

Dashboards, automated reporting, and data storytelling. Designing analytical models that make complex data intuitive and building visualizations that surface the insights that matter.

Python R SQL Quarto Visualization

How it works

1

Dimensional modeling

Star schema design with fact and dimension tables following Kimball methodology. Clean separations of measures and attributes.

2

Dashboard design

Purpose-built dashboards for different audiences: executive KPIs, operational monitoring, and self-service exploration.

3

Automated reporting

Parameterized reports that generate at scale. Replace manual data gathering with reproducible, templated outputs.

4

Analysis and storytelling

Cohort analysis, funnel breakdowns, segmentation, and trend analysis presented as clear narratives.

How this helps

Dashboards that load fast because they query well-modeled data, not raw tables.

Reports that generate themselves on schedule, freeing up hours of manual work.

Consistent metrics across teams because everyone queries the same models.

Insights presented in context, not just numbers on a screen.

retrain loop Feature Engineering Training Fit · Tune Evaluate ROC · Lift Deploy Cloud · Batch Monitor Drift · Alerts ML Lifecycle
03

Machine Learning

End-to-end ML development: from feature engineering and model training to production deployment and monitoring. Classification, propensity scoring, and forecasting applied to real business problems.

Python scikit-learn XGBoost Cloud ML pandas

How it works

1

Feature engineering

Build rich features from behavioral data, historical records, and cross-platform signals. Feature stores for training/inference consistency.

2

Model development

Train and evaluate models with proper validation: cross-validation, holdout sets, and business-relevant metrics beyond accuracy.

3

Production deployment

Deploy models on cloud infrastructure with batch inference pipelines. Predictions written back to the data warehouse for downstream use.

4

Monitoring and retraining

Track model performance over time. Detect drift, set up alerting, and schedule retraining when performance degrades.

How this helps

Scoring models that prioritize high-potential leads, letting teams focus where it matters.

Propensity models that predict behavior before it happens.

Automated inference pipelines that run daily without manual model execution.

Business decisions backed by data-driven predictions, not gut feeling.

</> React Astro TypeScript Tailwind Next.js Node.js Git
04

Web Development

Modern web applications with React, Astro, and TypeScript. From interactive data tools and dashboards to production-grade websites with strong UX and performance.

React Next.js Astro TypeScript Tailwind CSS AI-Assisted Dev

How it works

1

Architecture

Choose the right framework for the job: static-first with Astro, dynamic with React/Next.js, or hybrid approaches.

2

Component design

Reusable, typed components with clean APIs. Motion and interactivity where it adds value, not decoration.

3

Performance

Optimized bundles, lazy loading, proper caching, and Core Web Vitals as first-class concerns.

4

Deployment

CI/CD pipelines, automated builds, and hosting on modern platforms. Preview deployments for every change.

5

AI-augmented development

Leveraging AI coding assistants like Cursor, Windsurf, and Claude to accelerate development, generate boilerplate, and maintain code quality.

How this helps

Websites that feel fast and polished, reflecting the quality of your work.

Interactive data tools that let users explore insights without waiting for an analyst.

AI-assisted workflows that dramatically speed up development without sacrificing quality.

Production-grade code that's maintainable and built to last.

Trigger Schedule · Event Process Transform · Validate Database Warehouse Platform API · Sync Reports Dashboards Airflow CI/CD
05

Automation & Integration

Connecting systems and eliminating manual work. Data syncs, audience feeds, parameterized reporting at scale, and CI/CD pipelines that keep everything running.

Python Bash APIs CI/CD Airflow

How it works

1

Identify manual bottlenecks

Find the repetitive processes: manual data exports, copy-paste reporting, CSV uploads, and audience list generation.

2

Design the integration

Map source systems to destinations. Handle authentication, rate limits, data formats, and error recovery.

3

Build and orchestrate

Python scripts, API connectors, and shell automation tied together with scheduling and dependency management.

4

Monitor and maintain

Alerting on failures, logging for debugging, and documentation so others can understand and maintain the system.

How this helps

Hours of manual data gathering replaced by automated daily pipelines.

Downstream platforms always in sync with the latest data.

Fewer errors from manual copy-paste processes.

Processes that scale from a handful to hundreds without additional effort.