AI Engineer

Crogl
Fully Autonomous AI Team Member for Your SOC

Crogl is building AI-powered systems that help security teams investigate, understand, and respond to threats faster. We combine advances in large language models, agent architectures, and security expertise to create intelligent systems that solve real-world problems for security practitioners.

We’re looking for AI Engineers who are excited about building practical AI products. You’ll help design, evaluate, and improve agentic systems that operate in production, working closely with customers, engineers, and product teams to push the boundaries of what’s possible with modern AI.

This role is ideal for early-career and mid-level engineers who are passionate about AI, enjoy shipping products, and want to work at the forefront of LLMs and agent systems.

• Build LLM-powered features, workflows, and agentic systems that solve real customer problems.
• Design and implement evaluation frameworks to measure agent quality, reliability, and business impact.
• Create automated benchmarks, regression tests, and datasets for evaluating AI behavior.
• Investigate agent failures and develop systematic approaches to improve performance.
• Experiment with prompting, tool use, retrieval, memory, planning, and reasoning strategies.
• Build infrastructure that supports rapid experimentation, evaluation, deployment, and monitoring.
• Work closely with customers and internal teams to understand workflows and identify opportunities for AI automation.
• Contribute to engineering best practices for testing, observability, and production reliability.
• Stay current with advances in LLMs, agents, evaluation methodologies, and AI engineering.

Building agents is only half the challenge. Understanding whether they are actually improving is equally important. A significant portion of this role involves designing and maintaining evaluation systems for AI agents operating in real-world security workflows.

You’ll help answer questions like:

• Is the agent producing accurate investigations?
• Are changes to prompts, tools, or models actually improving outcomes?
• How do we detect regressions before customers experience them?
• Which failure modes matter most?
• How do we measure reliability, trustworthiness, and business impact?

You’ll build benchmarks, datasets, automated evaluations, regression testing pipelines, and observability systems that help us continuously improve agent performance.

• Strong programming skills, preferably in Python.
• Solid software engineering fundamentals, including testing, debugging, and system design.
• Experience building applications, projects, or products using LLMs and modern AI tools.
• Ability to design experiments, interpret results, and make data-driven decisions.
• Strong communication skills and willingness to collaborate across disciplines.
• Curiosity, ownership, and a desire to learn quickly.

• Experience building AI agents, copilots, or workflow automation systems.
• Experience designing evaluations, benchmarks, or testing frameworks for AI systems.
• Familiarity with OpenAI, Anthropic, Gemini, or open-source LLM ecosystems.
• Experience with retrieval systems, vector databases, and RAG architectures.
• Familiarity with Lang Graph, OpenAI Agents SDK, MCP, or similar agent frameworks.
• Experience with observability, tracing, and production monitoring for AI systems.
• Exposure to cybersecurity, security operations, or developer tooling.
• Open-source contributions, research projects, or personal AI products.

• Building evaluation suites that measure the quality of AI-driven security investigations.
• Creating automated regression tests to detect model and prompt regressions.
• Improving agent reasoning, planning, and tool-use capabilities.
• Developing datasets and benchmarks that reflect real customer workflows.
• Building systems that automatically identify and categorize agent failures.
• Designing feedback loops that continuously improve production agent performance.

The strongest candidates for this role are usually builders. You may have experience as a…