Nathan Aguirre
Senior Data Scientist building production ML for personalization, recommender systems, and customer behavior.
I help teams move from ambiguous product questions to deployed machine learning systems, including recommender models, churn and LTV forecasting, experimentation pipelines, monitoring, and stakeholder-ready insights.
Based in Berlin - Open to Senior/Staff DS roles - Available for select consulting engagements
Selected work
Systems built across streaming, ecommerce, insurance, and aerospace research.
A few examples of production ML, customer behavior modeling, experimentation, and senior IC ownership.
Crunchyroll
Personalization and retention ML for global streaming
Built and deployed personalization, recommendation, retention, LTV, and fraud models for a global streaming product. Work included recommender systems, subtitle-informed LLM/RAG recommendation approaches, campaign experimentation, PySpark/Python/SQL pipelines, KPI dashboards, and long-term model support.
- Improved recommendation diversity and relevance by 2x+.
- Supported double-digit churn reduction through predictive modeling and campaign experimentation.
- Helped prevent hundreds of thousands of monthly cancellation events.
Founding data scientist; helped establish DS tooling, event schemas, experimentation workflows, and project execution patterns.
Nordstrom
Product discovery and customer activation models
Built customer-facing product discovery features using topic modeling, clustering, recommender models, gradient boosting, and survival analysis. Work included product showcase generation, add-on recommendation improvements, customer activation modeling, and exploratory modeling over unstructured customer and product data.
- Shipped models into customer-facing personalization and activation workflows.
- Delivered material annualized business impact without disclosing exact financial values.
Connected discovery, activation, and product recommendations to practical ecommerce workflows.
Allstate
Forecasting and road safety analytics
Developed forecasting models for automotive sales signals and statistical methods for identifying hazardous road segments using telematics data. Work included validating road safety interventions and translating noisy behavioral data into actionable risk insights.
- Built ARIMA-style forecasting inputs.
- Applied geospatial and telematics analysis to road-risk questions.
Turned ambiguous risk and forecasting questions into usable analytical methods.
NASA
Applied ML and engineering research
Developed proof-of-concept approaches for clustering-based machine learning methods and contributed to engineering research related to shuttle repair methodologies.
- Worked on early applied ML prototypes.
- Contributed to repair methodology testing related to Atlantis and Discovery.
Early foundation in applied research, engineering judgment, and experimental methods.
Expertise
Organized around the problems I help teams solve.
The emphasis is production judgment: choosing the right modeling path, creating the measurement loop, and keeping the system useful after launch.
Personalization and recommender systems
Collaborative filtering, factorization machines, ranking systems, product and content discovery, topic modeling, clustering, and LLM/RAG-assisted recommendation approaches.
Customer behavior modeling
Churn prediction, LTV forecasting, activation modeling, retention campaigns, survival analysis, segmentation, and campaign targeting.
Experimentation and product measurement
A/B testing pipelines, KPI dashboards, campaign evaluation, model impact measurement, and stakeholder-facing reporting.
Production ML systems
PySpark, Python, SQL, Airflow, Databricks, MLflow, cloud deployment, monitoring, on-call support, and model maintenance.
Risk, anomaly, and fraud modeling
Fraud detection, anomaly detection, behavioral risk signals, and operational model monitoring.
Forecasting, time-series, and spatial analysis
Sales forecasting, ARIMA-style forecasting, telematics analysis, spatial risk methods, and intervention validation.
Senior IC leadership
Comfortable at the boundary between modeling, product strategy, and infrastructure.
My strongest work tends to start with an ambiguous product question and end with a deployed system, a measurement path, and a team that can operate it.
Early-team ownership
First data science hire on a team; helped establish tooling, workflows, event schemas, and reusable modeling infrastructure.
End-to-end execution
Built pipelines, models, dashboards, and monitoring paths rather than stopping at offline prototypes.
Cross-functional translation
Worked across product, marketing, engineering, and analytics to connect model outputs to product and business decisions.
Mentorship and project leadership
Led small project teams of data scientists and analysts while maintaining hands-on technical ownership.
Consulting
Available for select projects where senior data science judgment matters.
Consulting is secondary to full-time opportunities, but I am open to focused work where the scope fits the problem.
Available for select projects where senior data science judgment matters.
I occasionally work with teams that need experienced help scoping, building, or improving production ML systems, especially around personalization, recommender systems, churn and retention, experimentation, and customer behavior modeling.
Start a consulting conversation- Recommender system strategy and model review
- Churn, LTV, retention, and activation modeling
- A/B testing and measurement design
- Production ML architecture review
- Data science roadmap for early-stage teams
- Fractional senior DS advisory
Contact
Let's talk.
I am open to Senior/Staff Data Scientist roles and select consulting engagements in personalization, recommender systems, customer behavior modeling, and production ML.