Devoteam Cyber Trust | AI Engineer

Devoteam; Lisboa; On-site; EUR 70,000 - 100,000; Full time

Job summary

Devoteam seeks a product-driven engineer to design and evolve an autonomous pentesting platform. You will build agent-based systems, integrate LLMs, and orchestrate tool usage to automate significant parts of the pentesting lifecycle in production environments.

Responsibilities include architecture, task planning, state management, and validating agent behavior within real-world scenarios, ensuring reliability and security alignment with established WSTG practices.

Qualifications

• Strong proficiency in Python.
• Experience with APIs and distributed systems.
• Practical experience with LLMs (usage, integration, limitations).
• Understanding of agent-based systems.
• Ability to design complex, non-deterministic systems.
• Experience with task decomposition and planning.
• State management and execution continuity.
• Tool orchestration.
• Output validation and evidence handling.
• Ability to validate system behavior, not only code correctness.
• Experience with agent frameworks (Desirable).
• Background in Application Security (AppSec) and security testing automation (Desirable).
• Exposure to multiple LLM models and providers (Desirable).

Responsibilities

• Define and evolve the architecture of autonomous pentesting agents.
• Develop controlled execution pipelines (tasking, tool usage, feedback loops).
• Design and improve mechanisms for planning, state management, tool orchestration, and validation of agent execution.
• Integrate and optimize LLM-based systems within agent workflows.
• Define and validate tool-calling interfaces and integrations with pentesting tools.
• Ensure alignment with OWASP WSTG and other methodologies.
• Test and validate agent behavior in real-world scenarios.
• Identify, analyze, and mitigate system failures and edge cases.
• Contribute to internal standards, engineering practices, and design patterns.

Skills

Python

APIs

Distributed systems

LLMs

Agent-based systems

Complex systems design

Task decomposition

State management

Tool orchestration

Output validation

Job description

Job Description

Mission

Develop and evolve an autonomous pentesting platform based on agentic systems, integrated within the Offensive Security domain—specifically the Offensive Engineering and Innovation team—ensuring systems are effective, controllable, and capable of producing relevant and reliable outputs in real-world penetration testing scenarios.

Role Context

• This role sits within the Offensive Engineering and Innovation team, responsible for creating new technical capabilities that extend and scale offensive security services.
• The platform aims to automate significant parts of the pentesting lifecycle, aligned with methodologies such as the OWASP Web Security Testing Guide (WSTG), leveraging agents, LLMs, and integrations with existing security tooling.
• This is not an isolated experimental initiative. It is a production-oriented capability with direct application in delivery environments.

Responsibilities

• Define and evolve the architecture of autonomous pentesting agents
• Develop controlled execution pipelines (tasking, tool usage, feedback loops)
• Design and improve mechanisms for:
  • Planning
  • State management
  • Tool usage orchestration
  • Validation and control of agent execution
• Integrate and optimize LLM-based systems within agent workflows
• Define and validate tool-calling interfaces and integrations with pentesting tools
• Ensure alignment with established methodologies (e.g., OWASP WSTG)
• Test and validate agent behavior in real-world scenarios
• Identify, analyze, and mitigate system failures and edge cases
• Contribute to internal standards, engineering practices, and design patterns

Qualifications

Technical Skills:

Required

• Strong proficiency in Python
• Experience with APIs and distributed systems
• Practical experience with LLMs (usage, integration, limitations)
• Understanding of agent-based systems

Highly Important

• Ability to design complex, non-deterministic systems
• Experience with:
  • Task decomposition and planning
  • State management and execution continuity
  • Tool orchestration
  • Output validation and evidence handling
• Ability to validate system behavior, not only code correctness

Differentiators

• Experience with agent frameworks
• Background in Application Security (AppSec)
• Experience with security testing automation
• Exposure to multiple LLM models and providers

Soft Skills

• Strong critical thinking
• Ability to operate in imperfect and evolving systems
• Experimental and outcome-driven mindset
• Strong focus on control, reliability, and predictability
• High autonomy and ownership

Success Metrics

• Agents produce useful, actionable outputs
• Execution is predictable and controllable
• Low rate of unexpected or unsafe behaviors
• Effective integration into real pentesting workflows
• Continuous system evolution without uncontrolled complexity growth

Additional Information

The Devoteam Group works for equal opportunities, promoting its employees based on merit and actively fights against all forms of discrimination. We are convinced that diversity contributes to the creativity, dynamism and excellence of our organization. All of our vacancies are open to people with disabilities.