KarminaAI StudioAgents · Services · Training

JOBS · Collaboration by project

AI Agent Engineer · Tools, MCP & Multi-Agent Systems

Build agents that use tools, coordinate tasks, show trace and know when to stop.

The opportunity

Build agents who know how to act, explain themselves and stop.

We are looking for a profile of AI Agent Engineer to design systems that combine models, tools, status, knowledge and control points. The challenge is not to create a convincing chat, but to build a predictable, observable and useful agent within a business process.

Area
Agent Engineering
Experience
3+ years
Modality
Remote
Collaborating
By Project or Recurring

What are we looking for?

  • Experience with Python or TypeScript and APIs of generative models.
  • Knowledge of tool calling, structured outputs, context management, status and flows of several steps.
  • Experience with API Responses, Agents SDK or equivalent orchestration frameworks.
  • Practical understanding of MCP, tool contracts, authentication and permission limits.
  • Ability to design handoffs between agents or between agent and person.
  • Experience with tracking, logs, testing and failure management.

What are you gonna do?

  • Turn a business mission into an agent technical blueprint.
  • Design tools with inputs, outputs errors and explicit permissions.
  • Implement agents with memory limited to the purpose, recoverable state and useful traces.
  • Build multi-agent flows only when they provide a demonstrable advantage.
  • Integrate sources and actions without granting indiscriminate access to systems.
  • Collaborate with Evals to test behavior, regressions and boundary cases.
  • Document when the agent acts, when it recommends and when it must scale.

It will impress us especially.

  • A repository or demo with real tools, traces and reproducible tests.
  • Experience with user authorization, sandboxing, secret management or prevention of prompt injection.
  • Having reduced complexity by removing an unnecessary agent or tool.

What will you find?

Specific missions, delimited sources and defined success criteria before building. We value simple engineering, well-described tools and the ability to demonstrate why a system deserves to go from pilot to production.

How to introduce yourself

Attach your CV and add, if you want, a single link to your GitHub, technical portfolio or to the most rigorous agent you have built.

Talent

AI Agent Engineer · Tools, MCP & Multi-Agent Systems

This is the way to submit a candidacy or propose a professional collaboration. All fields are optional. Project and service inquiries are managed at hello@karmina.ai.

jobs@agenciakarmina.com