Master’s Research Internship (AI & LLM: From Theory to Application) – France

Job description details

Intro

Automated Prompt Optimization for Industrial Applications Based on LLMs

About the Position

As part of our efforts in AI research engineering, we are looking for a motivated final-year master’s intern (6 months) to work on automated prompt optimization for applications using large language models (LLMs).

Under the supervision of the Lead Research Engineer of the Security Lab, you will join the Prototyping, AI & Research team and actively contribute to projects aimed at making our systems more robust, scalable, and automated.

Hornetsecurity increasingly relies on LLMs for tasks such as classification, summarization, and explanation. Currently, these systems depend on manually crafted prompts, which:

• degrade over time as models evolve,
• require repeated manual adjustments,
• and lack scalability as the number of tasks grows.

This internship aims to address these challenges by developing automated prompt optimization pipelines, guided by performance metrics, suitable for industrial-scale deployment, while ensuring long-term maintainability and system quality.

The position is part of the AIESA project, with a strong research component and close collaboration with the applied AI teams.

At the end of the internship, the selected candidate will have the opportunity to continue in a 3-year CIFRE contract to pursue research and development within the project.

Your Job

The intern will be responsible for designing and evaluating an automated prompt optimization pipeline using frameworks such as DSPy. Key missions include:

• Exploring the state-of-the-art in prompting and optimization techniques,
• Implementing automated prompt optimization using open-source tools,
• Defining and tracking performance metrics (accuracy, recall, F1),
• Extending evaluation with text quality metrics (fluency, correctness, faithfulness, etc.),
• Integrating quality feedback through “LLM-as-a-judge” methods into the optimization loop.

Expected Outcomes

• A reproducible and automated pipeline, adaptable to multiple AIESA tasks and potentially extendable to other projects.
• A comparative study between human-crafted and optimized prompts.
• A written report establishing best practices for sustainable prompt engineering in production.

Your Profile

• Master’s student (or equivalent) in Computer Science, Artificial Intelligence, Machine Learning, or a related field, ideally with a research component.
• Strong interest in LLM prompt engineering and evaluation methodologies.
• Proficient in Python and modern machine learning tools.
• Research-oriented mindset, capable of:
  • critically reading and analyzing recent scientific papers,
  • formulating hypotheses and designing experiments to test them.
• Willingness to pursue a CIFRE PhD after the internship.
• Intellectually curious, autonomous, and rigorous, able to document work clearly and communicate results to both technical and non-technical audiences.
• Located in Paris or Lille.
• Fluent in English (written and spoken).

Your Benefits

• A dynamic, supportive, and learning-oriented environment.
• The opportunity to work on real and strategic cybersecurity challenges.
• 100% reimbursement of public transportation costs.
• Meal vouchers worth €10 per working day.
• CSE benefits & Student health insurance providing effective coverage from day one.
• 6 months internship contract duration.