P

Publicis Groupe

Runs advertising and digital transformation services

AI Engineer - Skills & Systèmes agentiques

AI Application EngineeringConfirmed live in the last 24 hours
Full-timeMontreal, QC, CanadaNo salary listed

About the Job

Requirements

  • Training or experience in software development or AI engineering.
  • Concrete experience building large language model solutions used by real users.
  • Proficiency in Python and APIs, and ability to read and modify production code.
  • Understanding of agentic systems, including tool calls, multi-agent orchestration, context management, and retrieval-augmented generation.
  • Experience with platforms such as OpenAI, Azure OpenAI, Anthropic Claude, Google Gemini, or Mistral AI.
  • A structured approach to evaluation, measuring quality rather than relying on subjective review.
  • Ability to facilitate workshops with business teams and turn unclear processes into clear specifications.
  • Clear communication in French and English.
  • Comfort working in a consulting context embedded with a client team.
  • Developer or data scientist background combined with business analysis ability.
  • Ability to investigate day-to-day processes with the people performing them and surface exceptions, implicit rules, and habitual decisions.
  • Ability to write clearly and structurally, explaining a task to someone without prior context.
  • Technical intuition about agent behavior and ability to diagnose failures across instructions, context, data, tools, or architecture.
  • Pragmatism, curiosity, and comfort working in a rapidly changing field.

Responsibilities

  • Work with business teams to document processes, decision rules, exceptions, and data used.
  • Determine which steps an agent can handle, which require human validation, and which should remain manual.
  • Translate data structures, including documents, tables, reference data, and APIs, into context usable by the agentic system.
  • Write and maintain agent instructions, skills, and tool definitions.
  • Structure skills to be modular, versioned, and reusable across use cases.
  • Define the context, data, and tools each agent receives and when it receives them.
  • Choose an approach appropriate to the need, such as instructions, data retrieval, tool calls, or code.
  • Make other agents, tools, and existing systems available to agents through skills.
  • Design flows between agents, including orchestration, context transfer, error handling, and human validation points.
  • Work with developers and the platform team to move solutions from prototype to production.
  • Build test sets from real cases provided by business teams.
  • Define quality criteria with business experts and implement automated evaluations, including LLM-as-a-judge.
  • Analyze failures such as hallucinations, incorrect tool use, or missing context, and address them at the appropriate level: instructions, data, tools, or architecture.
  • Implement guardrails and production monitoring, including traces and observability.
  • Document decisions and results so the client team can maintain the system.
  • Explain system capabilities, limitations, and trade-offs to client teams.
  • Train business teams to write and maintain their own skills where appropriate.
  • Share learnings with the MOOV team and contribute them to the client's roadmap.

Desired Qualifications

  • Experience with agent or orchestration frameworks such as LangGraph, Claude Agent SDK, OpenAI Agents SDK, or Semantic Kernel.
  • Experience with the Model Context Protocol or designing tools for agents.
  • Experience with evaluation and observability tools such as Langfuse, LangSmith, or Arize.
  • Knowledge of cloud environments such as Azure, AWS, or GCP, and CI/CD practices.
  • Experience in financial services or a regulated organization.

About the Company

Publicis Groupe

Runs advertising and digital transformation services

Publicis Groupe runs advertising, media buying and digital transformation services for companies that want to reach consumers, coordinate campaigns and improve their business systems. Its agency network includes creative and media specialists, while Publicis Sapient builds and modernizes the technology behind clients' operations. Epsilon adds customer data and marketing technology that connect brands with audiences and measure campaign results. Clients pay for consulting, campaign creation and execution, with production and media costs billed through alongside the group's service work. Founded in 1926, Publicis Groupe is headquartered in Paris, France.

Company size
10,001+
Company stage
IPO
Headquarters
Paris, France
Founded
1926

Recently Posted Jobs

Explore more opportunities at Publicis Groupe

See more jobs