
SQL and NoSQL modelling
Database design
When scattered files describe the same things, I design a common structure: tables or collections, relationships and integrity rules. I test it with representative examples before migration.

Independent consultant · Montpellier
I design databases, improve existing software and test practical uses of AI. For businesses and personal projects.
How I can help
A project can start with one clear problem: a query to improve, an import to make dependable, or an internal search to make more useful.
Indicative starting prices for business clients. Individuals receive a quote with the total price before any commitment.

SQL and NoSQL modelling
When scattered files describe the same things, I design a common structure: tables or collections, relationships and integrity rules. I test it with representative examples before migration.

SQL optimisation
I start with a slow query and a representative dataset. I record a baseline, compare changes in a test environment and document the trade-offs between reads, writes and storage.

Validation and governance
I design the intake for recurring files: field mapping, expected formats, duplicate detection and exceptions. A questionable row is never silently corrected; the reason stays visible.

Applied AI · NLP
I prototype a way to find passages in documents you authorise. Answers point back to sources and indicate when the collection cannot support an answer.

Machine learning
Before a rollout, I turn your criteria into test scenarios: typical cases, incomplete data and answers that need human review. The same evaluation grid makes two versions comparable.

Software quality
Within an authorised scope, I examine areas that are hard to maintain: access, dependencies, errors and test coverage. Recommendations are prioritised by impact and by what can be addressed within the work.
SQL and NoSQL modelling
When scattered files describe the same things, I design a common structure: tables or collections, relationships and integrity rules. I test it with representative examples before migration.
Indicative starting prices for business clients. Individuals receive a quote with the total price before any commitment.
SQL optimisation
I start with a slow query and a representative dataset. I record a baseline, compare changes in a test environment and document the trade-offs between reads, writes and storage.
Indicative starting prices for business clients. Individuals receive a quote with the total price before any commitment.
Validation and governance
I design the intake for recurring files: field mapping, expected formats, duplicate detection and exceptions. A questionable row is never silently corrected; the reason stays visible.
Indicative starting prices for business clients. Individuals receive a quote with the total price before any commitment.
Applied AI · NLP
I prototype a way to find passages in documents you authorise. Answers point back to sources and indicate when the collection cannot support an answer.
Indicative starting prices for business clients. Individuals receive a quote with the total price before any commitment.
Machine learning
Before a rollout, I turn your criteria into test scenarios: typical cases, incomplete data and answers that need human review. The same evaluation grid makes two versions comparable.
Indicative starting prices for business clients. Individuals receive a quote with the total price before any commitment.
Software quality
Within an authorised scope, I examine areas that are hard to maintain: access, dependencies, errors and test coverage. Recommendations are prioritised by impact and by what can be addressed within the work.
Indicative starting prices for business clients. Individuals receive a quote with the total price before any commitment.
Who it’s for
Reduce inconsistencies across files, measure a slow tool or prototype search across internal documentation.
Find material in personal archives, organise a collection or make a small tool you already use easier to maintain.

How I work
You show me a representative case. We agree which data is available, what is in scope and what success looks like.
I measure the starting point and propose a trial on an appropriate sample. You can verify the difference before expanding the work.
You receive the agreed deliverables, instructions to take them over and a clear record of what is still out of scope.
Yes. An outcome and an anonymised example are enough to start. You receive a proposed scope, total price and steps before committing.
Yes, once access and authorisation are agreed. I prefer a test environment and representative samples and do not request passwords through the form.
Yes. Projects can start with a remote discussion in French or English. We agree the format and availability together.
Get in touch
Tell me what you would like to improve and how you will judge the result. I will reply to agree the next step.
rafalrobert@inbox.eu+33 7 53 07 16 53Montpellier · remote projects · by appointment