Independent analytics engineer · Paris
Hi, I’m Adrienne. I help D2C and e‑commerce brands turn messy marketing data into decisions their teams can trust.
Available
Previously worked with
ba&sh PSG L’Oréal Groupe Homebox Nutrimuscle Easypara Dr. Pierre Ricaud
GRDF Kraaft Private Sport Shop
The problem
Sound familiar?
Three patterns I see in almost every engagement. If you recognise two of them, it’s usually worth a conversation.
01
Reconciling GA4 with the ad platforms eats half the week, and you’re still not sure which one is right.
Attribution
02
Your marketing team has stopped trusting the dashboards — so they go build their own in Sheets.
Trust
03
Your consent setup hasn’t been touched since GDPR, and you’re one CNIL letter away from a problem.
Compliance
None of these are unusual. All of them are fixable.
Services
Five ways I work with teams.
Start with an audit if you don’t know exactly what you need. That’s often the fastest way to find out.
01
Marketing Data Audit
Know exactly what’s broken — and what to fix first.
02
Tracking & Consent Setup
Tracking that works. Consent that complies.
03
Marketing Data Stack Build
A modern data stack built for how marketing teams actually work.
04
AI Agents for Data Teams
AI workflows that do the boring parts so your team can do the thinking.
05
Fractional Analytics Engineer
An analytics engineer on your team, without the hire.
Garden
From the garden.
A growing pile of working notes — dbt, BigQuery, marketing data, AI agents. Some are seedlings, some are evergreen. Worth a browse if you want to see how I think.
CLOUDSDK_CONFIG for Per-Project gcloud Isolation
How CLOUDSDK_CONFIG isolates all gcloud state per project — credentials, ADC files, active config — and why it's the missing piece for multi-client GCP work.
dbt Fusion Package Compatibility
How the dbt Fusion engine (v2.0) affects package compatibility — version bounds, manifest format changes, the Fusion badge, and how to prepare your project and packages for migration.
dlt and BigQuery Integration
How dlt loads data into BigQuery — the two loading strategies (streaming vs. GCS staging), the bigquery_adapter for partitioning and clustering, nested JSON normalization, and the metadata tables dlt creates.
Let’s talk.
Tell me what’s broken. I’ll reply within two working days with whether I can help — and if I can’t, I’ll point you somewhere useful.
Get in touch →