Data Platform Product Manager - 24 mois - Lyon - Permanent

Context

We collect a very large amount of behavioural, catalogue, subscription and advertising data, and we are currently migrating our data platform to Databricks with a data mesh model: each domain owns and publishes its own data products,

with near real time pipelines replacing batch.

We are looking for a Data Platform Product Manager to run the data platform as a product, and

then to build the client intelligence product on top of it.

The mission

You own the data platform as a product, for our product teams first and for our clients

next.

Two things make this different from a data platform role elsewhere. The platform is federated by

design, so almost everything you need is produced by squads you do not manage, and your job

is to make the mesh something people can find their way around and want to publish into, rather

than to fill it yourself. And the platform serves several broadcasters at once, across different

markets, business models and regulators, so every standard you set has to hold in more than

one place.

Concretely, you will:

● Set the vision and the plan first. Before any tool is built, we need a clear answer to

what the data product should offer, to whom, in what order, and why. That

written, validated vision is the first deliverable of this mission.

● Assess the current state. What data we already have and can trust, what is duplicated,

what is missing, and which of the questions we want answered are answerable today.

● Run the mesh as a product. Define what a data product has to carry before it counts as

one: a named owner, documentation, a schema, a freshness and quality threshold, and a

contract with its consumers. Publish that standard, and hold it.

● Make the data legible. Documentation, catalog and metric definitions, so that a person

who did not build a table and an agent reading it can both tell what is inside and whether

to trust it. This is the single largest gap today.

● Make domain ownership stick. Each domain owns the data it collects and why. Getting

five domains to publish, document and maintain their own data products is the hardest

part of this mission, and it is done by making publishing easy and the standard obvious,

not by escalating.

● Finish the migration and own what it costs. Complete the move to Databricks and to

near real time pipelines, and carry the platform cost per client as a number you report

rather than one you inherit.

● Build the internal product first. Product managers and squads should be able to see

the business impact of what they deliver, and the effect of client-side configuration

changes, without commissioning an analysis each time. Internal use is also how we prove

the numbers are right before a client ever sees them.

● Then open it to clients. A standard intelligence product our clients use to answer their

own business questions, built for their business users rather than only for their analysts,

and through which we come back to them with what to do about it. This includes working

with the business teams on whether intelligence sits inside the existing offer or is sold as

a module.

● Define how we know it worked. Adoption by product teams and by clients, decisions

actually taken on the product, analyst time released, the share of published data products

that carry an owner and documentation, and platform cost per client.

What is not in scope

● The content of other domains' data. You own the mesh, the standard and the tooling

that make domain ownership workable. You do not own what each domain collects, or

why. A single team cannot hold the meaning of millions of events a day across five

domains, and we are not going to ask you to.

● The API gateway and the developer documentation portal. These sit with

Infrastructure and with the documentation team, and they stay there until we have a

packaged product to sell with usage tracking that supports billing.

Profile we are looking for

Required

● Senior product manager with data product management experience. At least 7 years

in product, minimum 3 of them owning data, analytics or intelligence products rather than

consuming them.

● You have run a data platform as a product. You held the vision, the standard and an

adoption number, and the engineers who built it worked with you rather than for you.

● You have worked in a federated or data mesh model and can tell us plainly what made

domain ownership stick and what did not.

● You have moved a documentation, coverage or data quality number through teams

that did not report to you. Tell us what it went from, to what, and over how long, and

how you got those teams to care.

● You have built data products before. An analytics or intelligence product that people

used to make business decisions.

● B2B context. Products used by business customers or by internal operators, ideally

multi-tenant.

● Comfortable in a modern data stack. Warehouse-native analytics (Databricks,

Snowflake or equivalent), event data, semantic layers, catalog and governance tooling, BI

tooling, data quality.

● Visionary, and able to make the vision concrete. You can describe a data product that

does not exist yet, and then break it into increments that ship.

● Systems thinking. You reason about second-order effects, and you name dependencies

and prerequisites early while proposing the way forward.

● AI-native. You use AI tools in your own daily practice, and you have a considered view on

where AI belongs on top of data and where the data is not yet clean enough to carry it.

● Fluent English. Working language across our international teams and clients.

Appreciated, not required

● Data products built under a regulatory or consent constraint.

● Experience preparing data for AI consumption: catalog quality for retrieval, semantic

definitions an agent can read, evaluation sets.

● Experience owning a cloud data cost line.

Where the role sits

You report to the VP Product, Platform as a Product. You work with the Data Platform squad, the

Tracking squad, the data chapter, our engineering leadership, and the product managers of the

five other domains who publish into the mesh.