Services
Data and artificial intelligence services.
I work with companies and teams that need to prepare their data, understand it, automate processes, or develop models and AI systems.
I can take on a specific need, or join a wider project from the early stages through to a solution people can use.
Data preparation and cleaning
The data comes from scattered sources, with duplicates, missing values, formats that do not fit, inconsistencies, or records that do not add up.
I can take on cleaning, normalisation, transformation, integration, reconciliation, and validation. It can be the whole commission, or a phase with its own scope inside a larger project.
Possible commissions
- Bring scattered sources into one set that can be used.
- Prepare a dataset for an analysis or a model.
- Clean a catalogue.
- Remove duplicates and validate records.
- Leave the information ready for an AI system.
Data analytics and BI
Indicators do not match, reports are built by hand, there is little visibility, or the analysis is hard to interpret.
I do exploratory and statistical analysis, look at correlations and how variables relate, define KPIs, and build dashboards, including Power BI, plus reports that refresh on their own.
Possible commissions
- Campaign analysis.
- Commercial performance.
- Descriptive segmentation.
- Operational monitoring.
- Recurring reports.
Data science and machine learning
You need to anticipate an outcome, separate groups, see what falls outside the normal pattern, or compare ways of modelling.
I build prediction and classification models, group similar cases (clustering), forecast a series over time, and detect fraud or anomalies. When place is part of the question, the analysis can be geospatial. The work includes choosing variables, seeing whether they explain one another (multicollinearity), keeping a model from fitting the past too closely (regularisation), and judging whether the result is useful.
Possible commissions
- Predicting churn.
- Forecasting demand.
- Customer segmentation.
- Spotting suspicious transactions.
- Analysis aimed at inventory.
Data engineering
Processes are slow or brittle, work is still manual, a migration is complex, or you cannot trust what a data flow produces.
I design loads that extract, transform, and land data where it is used (ETL/ELT), pipelines, migrations and modernisation, automation, processing, and quality checks. Depending on the case I use SQL, Python, PySpark, Databricks, or SSIS.
Possible commissions
- Move SQL Server processes to PySpark.
- Automate loads.
- Improve pipelines that already run.
- Reconcile results.
- Prepare flows for analytics or machine learning.
AI systems
You need to automate work on documents, retrieve information, put a model inside an application, judge answers, or see whether a new version is worse than the last.
I build applications that retrieve passages and answer with a language model (RAG), language work (NLP), document reading (OCR), and vision models. I also evaluate RAG and LLM systems, compare models, set up regression tests, and improve a system that already runs.
Possible commissions
- Assistants over documentation.
- Information extraction.
- Evaluation of retrieval and of the answers.
- Model comparisons.
- Checks before updating an AI system.