Senior AI Platform Architect

Context

A leading full-service vehicle leasing and mobility services company, part of a major banking group, is building its AI Factory: an industrial organisation responsible for designing, producing, and exposing artificial intelligence capabilities to its entities across approximately thirty countries.

The purpose is not experimentation but to transform a set of AI solutions—currently delivered unevenly across countries and business lines—into a catalogue of products, platforms, frameworks, and services that are genuinely consumable: documented, backed by service levels, subscribable through the Group marketplace, charged on usage, and traceable end to end.

The technical scope currently covers data science and generative AI platforms (Dataiku, Domino), an intelligent document processing chain built on ABBYY FlexiCapture and its trajectory towards ABBYY Vantage, LLM as a Service and RAG as a Service offerings, conversational agents (RASA, Genesys, Salesforce), an API exposure gateway (Apigee), a robotic process automation platform, and MLOps, DataOps, and DevSecOps frameworks aligned with the Group DevOps toolchain. Deployments run on the Group cloud, in the dMZR zone, with a requirement for as-code industrialisation.

The AI Factory operates within a structured ecosystem: Enterprise Architecture, which sets standards and validates solutions; the Group DevOps and Data centres of expertise; Group IT, which produces part of the offers consumed; and the Group marketplace, the subscription and provisioning channel.

The assignment takes place during a build phase: the capabilities exist technically but are not yet industrialised nor exposed as offers. The incumbent will contribute directly to that transition to scale.

Missions

• Define and document the technical and functional architectures of the AI Factory's products, platforms, frameworks, and services.

• Submit those architectures to Enterprise Architecture for validation and integration into the application landscape.

• Ensure the integration of solutions into the Group cloud, working with production teams and the DevOps centres of expertise.

• Compose architectures by favouring reuse of building blocks available on the Group marketplace, and document what the AI Factory adds whenever it extends a Group offer.

• Support operations teams in the as-code implementation of those architectures.

• Feed the offer catalogue entries with everything architecture-related: technical content, dependencies, prerequisites, limits, and known weaknesses.

Scope boundary:
The AI Factory architect designs and submits; Enterprise Architecture validates. The incumbent substitutes for neither function.

Technical directions are prepared together with the engineers who will implement them, and those engineers attend the forums where such decisions are made. No architecture is arbitrated without them.

The incumbent is not responsible for project management or for prioritising work, which remain with the Product Owner and the pole manager.

Main activities

Architecture design:

• Produce the technical and functional architecture dossiers for the catalogue offers.

• Model the flows: document ingestion, processing orchestration, API exposure, event propagation, object storage, and evidential archiving.

• Define reusable reference patterns: API-first exposure, multi-service ingestion gateway, RAG architectures, model training and deployment chains.

• Arbitrate component choices — vector store, storage engine, runtime — on the basis of written and compared criteria.

Submission and compliance:

• Prepare and defend dossiers before Enterprise Architecture and Group architecture bodies.

• Ensure compliance with Group standards: security, hosting, data localisation, evidential archiving, and traceability of AI usage.

• Maintain the application and architecture repositories for the scope.

Cloud integration and production:

• Design deployment targets on the Group cloud and in the dMZR zone.

• Work with the DevOps and DevExp centres of expertise on integration with the Group CI/CD toolchain.

• Specify observability, disaster recovery, scalability, and load-handling requirements, and verify that they are achievable before any commitment is made.

Group marketplace and reuse:

• Map the building blocks available on the Group marketplace and identify those covering all or part of an AI Factory need.

• Systematically document, for any Group offer extended locally, what the AI Factory adds and why — this justification conditions entry into the catalogue.

• Prepare the dossiers required to expose an AI Factory offer on the marketplace: technical description, subscription prerequisites, orchestration delivering the digital artefact.

As-code enablement:

• Translate architectures into reusable infrastructure-as-code modules.

• Support operations teams in their implementation and review the infrastructure code produced.

• Contribute to the templates of the MLOps, DataOps, and DevSecOps frameworks in scope.

Tools & Environment

• Data science and generative AI platforms: Dataiku, Domino or market equivalents.

• Intelligent document processing: ABBYY FlexiCapture or Vantage, or equivalents (Kofax, Hyland, Azure Document Intelligence).

• LLM and RAG services: inference architectures, vector stores (Elasticsearch, native vector engines), embedding management.

• Conversational agents: RASA, or agents embedded in Genesys and Salesforce.

• API management and exposure: Apigee or equivalent.

• Data flows: Kafka, change data capture, object storage, electronic archiving.

• Cloud and industrialisation: Azure and enterprise private cloud; containerisation and orchestration (Docker, Kubernetes).

• Infrastructure as code: Terraform, Ansible or equivalent.

• Industrialised CI/CD toolchains, GitLab, MLflow, and MLOps tooling.

Working Conditions

• Contract details not specified.

• Location: On site at Rueil-Malmaison, 4 days per week.

• Start date and duration not specified.

• Compensation/rate not specified.

1. Minimum 8 years' experience in technical architecture, including at least 3 years in data or artificial intelligence platform architecture
2. At least one experience taking a data or AI platform into production in a regulated environment (banking, insurance, healthcare, public sector) validated by an enterprise architecture body
3. Proven command of at least one market data science or generative AI platform, evidenced by a verifiable reference
4. Command of infrastructure as code and CI/CD toolchains, evidenced by a verifiable reference
5. Demonstrated experience integrating a solution into an imposed enterprise DevOps toolchain
6. Design of distributed architectures in a banking or regulated large-enterprise environment
7. Command of exposure patterns: API-first, event-driven architectures, microservices, API management
8. Functional modelling and ability to translate a business need into a technical service reusable by several consumers
9. Experience with technologies such as Dataiku, Domino, ABBYY FlexiCapture or equivalents, LLM and RAG services, vector stores (e.g., Elasticsearch), RASA, Apigee, Kafka, object storage, and electronic archiving
10. Experience with Azure and enterprise private cloud, containerisation and orchestration (Docker, Kubernetes)
11. Experience with Terraform, Ansible or equivalent for infrastructure as code
12. Experience with industrialised CI/CD toolchains, GitLab, MLflow and MLOps tooling
13. Ability to collaborate closely with engineers during solution building rather than handing over finalized solutions
14. Ability to produce and maintain genuinely usable documentation
15. Ability to defend dossiers before architecture bodies and accept arbitration
16. Fluent French (written and spoken) and professional English
17. Availability to work on site at Rueil-Malmaison 4 days per week