Its clients include more than 450 marketplaces, both B2C and B2B, and a network of over 100k third-party sellers. Brands such as Macy's, Decathlon, Carrefour, Asos and Airbus Helicopters use its products.
The company's 5 values are working hard together, getting things done, going beyond what is expected, learning through expertise, and making clients successful.
The Mirakl Assistant team is new and works like a start-up inside a scale-up. It is led by Remi Takase, Senior Data Science Manager, and sits within a data organisation of data scientists, research scientists, data engineers and SREs, built on a machine learning and data platform.
The team is building a conversational AI assistant that helps users learn the platform, understand their business and grow it. It is moving from answering questions to carrying out tasks on the user's behalf.
AI is central to the company's product strategy, from catalogue transformation and content generation to AI agents and agentic commerce.
Responsibilities
Design, build and scale the assistant's core parts, including RAG, LLM orchestration, tool use, memory and context engineering, taking them from rapid prototype to robust production features.
Own the full production lifecycle of the assistant: automated data pipelines, CI/CD, deployment, load testing, latency and resilience.
Lead technical discovery and feasibility studies for new capabilities and integrations across the product range.
Raise quality through rigorous offline and online evaluation, data-driven experiments and continuous iteration.
Work with software, data engineering, SRE and product teams to evolve the agentic foundations and turn user needs into capabilities.
Requirements
Have 5+ years as a machine learning engineer with strong software engineering skills, or as a software or data engineer with hands-on AI and ML experience.
Have a proven record of deploying AI/ML systems in production products, with an automation-first approach.
Have proven experience building AI agents with LLMs, RAG, MCP and related technologies, including architectural design.
Have a strong background in data-driven iteration, from experimentation through to evaluation.
Have a solid grounding in machine learning algorithms and fundamentals.
Be proficient in Python software engineering.
Have experience with LLM orchestration and agentic systems design (RAG, MCP, LangChain, etc.).
Be rigorous about metrics-driven evaluation and experimentation.
Have a '0 to 1' mindset: thrive in ambiguity and take an idea from exploration to production.
Understand end users' pain points and balance technical excellence with customer impact.
Communicate clearly and collaborate well across functions.
Be fluent in English and French.
Frontend development with TypeScript and React is a plus.
The stack you will work with: Python, SQL and TypeScript; FastAPI, LangChain, OpenSearch, Kafka and PostgreSQL; React with micro-frontend architecture; LLMs such as OpenAI and Anthropic models; Spark, Delta Lake and Databricks Workflows; Langfuse and MLflow; Docker, Kubernetes, GitHub Actions, ArgoCD, AWS and GCP.
Conditions
This is a permanent contract based in Paris or Bordeaux, with up to 1 day of remote work per week.
You are encouraged to apply even if you do not match every requirement.
AI tools may help screen applications and assess responses, but humans make all final hiring decisions.
The company welcomes people from diverse backgrounds and experiences.
Good
Your chances
Still hiring, not crowded yet, and you'd be among the first.
16 checks run
6 good signs
1 red flag
Still hiring?
10 checks
Actively hiring
StaleHiring
In its favour2
Still on the company's own careers site, checked 1 h agoModerate evidence
Found in the last 48 hours, before the big job boardsModerate evidence
Against it1
Posted 4 times at once for the same place: mass postings are often a pipeline, not one seatSlight evidence
How crowded?
6 checks
Low
QuietCrowded
In its favour4
In its Early Window: not on the big job boards yetStrong evidence
Senior level: far fewer people qualifySlight evidence
Asks for LangChain, which only 2% of open roles doSlight evidence
1 moreFewer
Asks for French: far fewer people can applySlight evidence
?
Fits Me
How well does this role fit you?
Answer a few questions or drop your CV, and every role gets a fit score with the reasons, this one first.