Architecte Technologies IA – Large Language Models (LLM) chaîne logistique et ingénierie
Paris
Posted 2d ago
Job Location
Paris
Tech Stack
Remote Work Policy
On-site
Categories
AI Agent Engineer
About the job
As an operational AI/LLM Architect, you will be at the forefront of designing and developing advanced AI systems that power modern businesses. This is a highly technical and hands-on role where you will spend most of your time architecting and engineering real-world AI solutions across traditional machine learning, generative AI, and agentic systems within active client engagements. You will translate business needs into concrete architectural decisions, selecting appropriate design patterns, evaluating and benchmarking technical frameworks, assembling reusable components, and making technology choices that balance innovation with enterprise-grade reliability. You will design and develop AI agent architectures, including multi-agent orchestration, tool use, skills, and memory systems, while working directly with foundation models through fine-tuning, Retrieval-Augmented Generation (RAG), and custom model integration. A significant part of your work will also involve designing and implementing the AI Context Layer that makes these systems truly intelligent in production. You will connect enterprise knowledge bases, structured and unstructured data sources, and specific business content to ensure generated responses are reliable, relevant, and aligned with client objectives. You will design and validate systems against enterprise non-functional requirements, including security, observability, governance, performance, and scalability. A key deliverable of this role is the production of concrete architectural and engineering artifacts. This means you will develop and own software components as a practitioner, building, integrating, and testing AI system modules, while producing detailed architecture documents such as Architecture Decision Records (ADR), component diagrams, data flow diagrams, and integration specifications that will serve as a reference for engineering teams. You will collaborate with multidisciplinary teams of Data Engineers, ML Engineers, and application developers. This position offers an excellent opportunity to develop deep expertise across the entire AI architecture stack, refine your engineering instincts on complex and concrete problems, and build a career path towards a Lead Architect or Principal Architect role.
Responsibilities
- Design, develop, and autonomously deliver software components across the AI architecture, taking end-to-end ownership of design, development, integration, and testing.
- Design and develop AI agent architectures, including individual agents, their prompts, tools, and skills, multi-agent orchestration, and memory systems, while making informed architectural pattern and technology choices.
- Design and implement agent orchestration models capable of managing task handoffs, communication, state management, and error recovery, then validate them through operational prototypes.
- Evaluate multiple design options and technical approaches to make justified architectural decisions, balancing features, cost, performance, and enterprise-grade reliability.
- Design, develop, and execute evaluation strategies to measure the quality of agents and systems on criteria such as accuracy, relevance, and faithfulness, then translate these results into design improvements.
- Architect and implement foundation model integration, selecting the most suitable models, invocation methods, and customization approaches (fine-tuning, RAG, custom integrations) based on performance, cost, and capability trade-offs.
- Design and develop model adaptation and fine-tuning pipelines, applying a strong understanding of Transformer-based architectures to optimize model selection and performance.
- Design and develop the AI Context Layer, including the design of context graphs and ingestion pipelines capable of analyzing, chunking, enriching, and indexing structured or unstructured enterprise content, as well as retrieval components to ground AI responses in client knowledge.
- Develop embedding generation, vector storage mechanisms, and retrieval components (semantic, hybrid, reranking) within complete RAG pipelines connected to enterprise data systems.
- Design and implement context assembly mechanisms and memory systems to manage prompts, context windows, and conversational state to produce accurate and contextualized responses.
- Identify, design, and develop reusable components and solution patterns that accelerate deployments and can be industrialized across multiple engagements.
- Design optimized solutions.
Requirements
- Translate business needs into concrete architectural decisions.
- Select appropriate design patterns.
- Evaluate and benchmark technical frameworks.
- Assemble reusable components.
- Make technology choices that balance innovation and enterprise-grade reliability.
- Design and develop AI agent architectures, including multi-agent orchestration, tool use, skills, and memory systems.
- Work directly with foundation models via fine-tuning, RAG, and custom model integration.
- Design and implement the AI Context Layer.
- Connect enterprise knowledge bases, structured and unstructured data sources, and specific business content.
- Design and validate systems against enterprise non-functional requirements (security, observability, governance, performance, scalability).
- Produce concrete architectural and engineering artifacts (ADRs, component diagrams, data flow diagrams, integration specifications).
- Collaborate with multidisciplinary teams (Data Engineers, ML Engineers, application developers).
- Develop expertise across the AI architecture stack.
- Refine engineering instincts on complex, concrete problems.
- Build a career path towards Lead Architect or Principal Architect.
- Design and develop software components end-to-end.
- Design and develop agent orchestration models.
- Evaluate design options and technical approaches.
- Design, develop, and execute evaluation strategies for AI agents and systems.
- Architect and implement foundation model integration.
- Design and develop model adaptation and fine-tuning pipelines.
- Develop embedding generation, vector storage mechanisms, and retrieval components.
- Design and implement context assembly mechanisms and memory systems.
- Identify, design, and develop reusable components and solution patterns.