Senior AI Engineer

Decathlon aims to become the best sports digital platform and the open ecosystem of the world. We want to enable customers to experience Decathlon through many local, sport-focused experiences by connecting many third-party players and services, securely and efficiently.

Our digital teams based in Lille and Paris (and beyond…), comprising more than 5,000 collaborators, are united in designing and developing digital products with the goal of always offering the best value to our users. Present in more than 70 countries, Decathlon is committed to innovation, sustainability and customer satisfaction.

Job Description

Our PRISM team (Supply Chain) is recruiting an AI Engineer based in Paris. As an AI Engineer dedicated to deploying agents, you will be at the heart of a tight-knit technical team. You will play a critical role in the industrialisation, deployment and monitoring of our agentic capabilities. You will work closely with the digital teams (Data, Product, Engineering) to architect the deployment infrastructure, design intelligence routing and build a secure infrastructure allowing our agents to interact natively with our enterprise data and our legacy systems

Responsibilities

• Design and deploy the solid foundations that will allow our AI agents to be hosted, to run and to interact smoothly with our systems when going into production.
• Design robust pipelines enabling asynchronous, multi-step agentic execution (e.g. transactional ingestion and write flows), implementing distributed resilience patterns.
• Integrate agentic orchestration SDKs (notably Google Vertex AI Agent SDK / ADK), working around the "black box" limitations of managed frameworks. 
• Orchestrate the agents in production efficiently while ensuring proper handling of information processing, routing to the most suitable models, and optimal cost and token usage management. Guarantee and ensure the quality and speed of execution of the tasks to be carried out. 
• Strictly frame the scope of action of each agent. You ensure they access only the data and actions they are formally authorised for, in compliance with our privacy rules, ensuring data security. 
• Build advanced performance measurement and tracking pipelines to trace the agent's reasoning and decisions end to end in order to debug logic in production.
• Work with the team to define relevant and consistent data evaluation metrics aligned with the specific functional metrics defined.

Know-How

• Design of robust backends (Python), operational mastery of containerisation (Docker, Kubernetes) and pipelines (CI/CD, automated testing, observability, log management) to ensure robust production deployments.
• Production-focused Python expertise, mastery of deployment platforms (Vertex AI Agent Builder, Gemini Enterprise Agent Platform) and client/server integration protocols such as the Model Context Protocol (MCP).
• Experience optimising context windows for massive frontier models. Routing requests to the models suited to the envisaged solutions
• Ability to build test frameworks for LLMs (LLM-as-a-Judge, Golden Datasets), mastery of deterministic metrics and strict validation of output formats (JSON schema enforcement).
• Hands-on experience with Agentic RAG, Semantic Caching to reduce latency and API costs, or the use of prompt optimisation frameworks (e.g. DSPy).
• Applied experience in supply chain, retail, or model deployment via Databricks / MLFlow.

Soft Skills

• Excellent communication and collaboration skills.
• Ability to work effectively in a dynamic and fast-paced environment.
• Strong problem-solving skills and attention to detail.
• Adaptability and willingness to learn new technologies and methodologies, and to share them. 
• Ability to translate complex technical concepts into terms understandable to non-technical stakeholders.
• Ability to understand user challenges and associated needs

Qualifications

• Senior profile: You have solid experience (Machine Learning Engineering) with a strong orientation towards putting infrastructure into production.
• Ability to leverage managed solutions (Vertex AI) while knowing how to go into the details to fix distributed architectures, develop custom connectors and work around framework limitations.
• Proven ability to thrive in shifting technical environments
• Operational fluency on GCP and understanding of containerised environments (Docker, Kubernetes) to autonomously deploy the agentic layer.

Technical Environment

• Agent orchestration: Gemini Enterprise Agent Platform, Vertex AI Agent Builder
• Infrastructure & Cloud: GCP (Google Cloud Platform), Docker, Kubernetes, CI/CD pipelines.
• AI & Integration protocols: Model Context Protocol (MCP), LLMs (Gemini), LLM routing & memory management.
• Security & Access: Google Cloud Model Armor.
• Execution Engine: Databricks, AWS EKS, Sagemaker
• Payload: Python, Spark, Scikit-Learn, Tensorflow / Pytorch, Pyspark 
• CICD: Github Actions
• Serving: Docker, Protobuf, gRPC
• Model registry, Model Tracking: MLFlow
• Orchestration:  Airflow
• Documentation / code: Git, Confluence 
• Data Visualisation: Tableau

WHAT WE OFFER YOU

• 2 days of remote working per week;
• Equipment provided in line with your missions and our societal commitments (Mac, Windows, or Chromebooks);
• A local project team within a global network (possibility of an international career);
• Skills development and support (variety of projects, technical certifications from the first year, internal and external training, etc.) 
• Remuneration package (employee participation in company performance, monthly/quarterly bonuses).

Joining Decathlon means being part of a dynamic team passionate about sport, innovation and having a positive impact on the lives of our customers. If you are enthusiastic about leveraging data science to optimise supply chain processes and drive business growth, we invite you to apply and become part of our adventure.