The deployment of AI agents based on Large Language Models (LLMs) opens up interesting prospects for the design of new so-called agentic systems [1]. An agentic system is characterized by an interface that uses natural language as the primary means of interaction with the user. Using an LLM as the engine of such a system makes it possible to translate user needs into commands that can be directly executed by any system (computing, electronic, etc.) that provides a server based on the Model Context Protocol (MCP) [2].

The objective of this internship is to design and implement an Agentic AI Web Platform that can be used and managed through natural language instructions. The platform will include several components, such as managing web servers, hosting web applications, building web applications, and handling the entire software chain of web technologies. All of these elements should be operated using natural language instructions (in English or French). The main technologies to be used are LLMs and MCP. In addition, we aim to build this platform using code generation tools, thereby extending the possibilities available to users for building new applications. Security issues are an important part of this project. Giving LLMs advanced privileges on web servers may induce big security threats that need to be studied and prevented before the deployment of the solution. A diagram of the platform is shown in the figure below:

This internship focuses on a very recent technology whose scope and limitations are not yet fully understood. The objectives of the internship are to explore and apply this technology in order to achieve the project’s goals. Depending on the intern’s skills, ability to grasp these new technologies, and progress during the internship, the objectives may be adjusted. The ultimate goal is to create a system that can be used by anyone capable of communicating in natural language (English or French), without requiring prior technical knowledge of its implementation.

The internship will take place within the IDEA service (Information Technology: Development, Operation, Support) at LAAS-CNRS and is intended for a final-year engineering student or a Master’s 2 student. The candidate should have a strong interest in LLM-based applications and agentic systems in general. Programming skills in Python are required to successfully carry out this work, and thorough knowledge of web hosting and web application technologies is essential.

Application:

Candidates must submit their application including a cover letter, CV, and transcripts from their higher education studies. A letter of recommendation is an asset for the application.

Contact :

References

[1] Agentic AI - https://en.wikipedia.org/wiki/Agentic_AI

[2] Model Context Protocol - https://modelcontextprotocol.io/docs/getting-started/intro