Agentic AI Engineer (Agents, Skills, Kilo Code) - Regression Testing in Banking

Project overview

We are building a focused team for a major banking client within its CIB (Corporate and Investment Bank). An AI engineer and two process experts will work together. Quick start.

The mission focuses on regression testing of market risk calculation applications. The two experts will analyze and redefine the process with the bank's teams. You will transform this work into agents: which agent to build, for which task, with what knowledge.

Your main missions:

What the mission offers you:

Profile sought

AI Engineer with practical experience in missions involving:

You know how to test your agents and work within the rules of a development cycle (SDLC) in a constrained environment. You enjoy transforming expert input into solutions. Experience in software testing or finance is a plus.

Work environment

Mission in Paris, on-site alongside the teams, within the IT department of a corporate and investment bank.

Sensitive data: everything is built within the bank's secure environment. Initial phase is focused, with subsequent phases depending on results.

• Translate the processes optimized by the experts into concrete AI agents.
• Create the agents and code the skills and knowledge they need, on Kilo Code, the tool chosen by the bank.
• Define with the experts the knowledge and business rules to be given to each agent.
• Test and measure your agents, compare their results to those of the teams.
• Adjust your agents through short iterations, using an agile approach, with the bank's testing teams and AI teams.
• Agentic AI on a real-world case, with measured results.
• A partnership with experts who will guide you on the business: investment banking knowledge is not required.
• A place in a focused team where your technical choices matter.
• Creating agents and skills (instructions, scripts, tools, and packaged knowledge);
• A code assistant or agentic tool: Kilo Code, or equivalent like Claude Code;
• Providing knowledge to agents (RAG, MCP);
• Python.

Project details

• Start: ASAP
• Duration: 1 an renouvelable
• Location: Paris, France, FR
• Languages: French