AI Cybersecurity Engineer - LLMs & Agent Pipelines

Crane Venture Partners; Paris; Sur place; EUR 44 000 - 54 000; Plein temps

Avantages offerts par ce poste

Health insurance

Mental wellness app

Meal vouchers

Paid time off

Résumé du poste

HarfangLab, a French cybersecurity scale-up, is seeking an AI engineer to join the Artificial Intelligence team in Paris. You will help design, develop, and deploy advanced threat investigation capabilities, leveraging a Large Language Model to assist analysts and automate investigations.

You will work on prototype-to-production features, evaluating models, and integrating solutions into HarfangLab products while collaborating with a small, autonomous team.

Qualifications

• Proficient in reading Python code and understanding AI systems.
• Strong knowledge of how LLMs work: data gathering, prompting, fine-tuning, inference, attention, tokenization.
• Hands-on experience with LLMs via APIs or local deployments and building apps like RAG, tools, agents, or chatbots.

Responsabilités

• Build investigation agent pipelines (planning, tool calling, synthesis).
• Design and maintain an evaluation framework to measure performance and metrics.
• Evaluate foundation models against use cases and document trade-offs between cost, latency, and quality.
• Industrialize developments with testing, packaging, documentation, and CI/CD practices.
• Integrate solutions into the product, deploy and monitor in production.

Connaissances

Python

LLMs

APIs

RAG systems

Tool-calling workflows

Outils

LangGraph

PydanticAI

Haystack AI

PyTorch

Description du poste

HarfangLab, a French cybersecurity scale-up, is seeking an AI engineer to join the Artificial Intelligence team in Paris. You will help design, develop, and deploy advanced threat investigation capabilities, leveraging a Large Language Model to assist analysts and automate investigations.

You will work on prototype-to-production features, evaluating models, and integrating solutions into HarfangLab products while collaborating with a small, autonomous team.