Machine learning

Models that predict something your business acts on, measured against a simple baseline so you can see what they add.

Typical deliverables

• Demand, revenue and churn forecasts with prediction intervals
• Classification, ranking and recommendation models
• Feature pipelines, model evaluation reports and monitoring

Python · scikit-learn · XGBoost · PyTorch · MLflow

ETL and data engineering

Data cleaning, analytics engineering and pipelines that get data from where it lives to where it is useful, on schedule, with failures that are easy to find.

Typical deliverables

• Batch and incremental pipelines from APIs, files and databases
• Warehouse models with tests and documentation
• Migrations from slow scripts and spreadsheets to fast, maintainable code

Polars · SQL · dbt · Airflow · DuckDB

Data visualization

Charts and dashboards designed around the questions people ask, without decoration that gets in the way.

Typical deliverables

• Executive and operational dashboards
• Metric definitions that everyone in the business shares
• Publication-quality charts and automated reports

Quarto · Plotly · Altair · Power BI · Tableau

Applied AI

LLM features that do a specific job well, tested on your own examples before anyone relies on them.

Typical deliverables

• Search and question answering over internal documents
• Extraction of structured data from text, PDFs and email
• Evaluation sets that track quality as models and prompts change

Python · LLM APIs · vector search · evaluation harnesses

Statistics and experimentation

Careful analysis for the questions where a wrong answer is expensive, with the range of plausible outcomes stated clearly.

Typical deliverables

• A/B test design, sample size planning and readouts
• Causal and regression analysis on observational data
• Reviews of existing analyses and metrics

Python · statsmodels · Bayesian modeling

Training

Practical workshops built around your team’s own data problems, so the skills carry straight into daily work.

Typical deliverables

• Polars and modern Python data analysis
• SQL and data modeling for analysts
• Code review and mentoring during a project

See the books and courses for a sense of the teaching style.

Workshops · course material · mentoring