MSc Intership / Thesis: Secure LLM Agents for Interactive Urban Digital Twins

  • 6 to 9 months
  • The Hague
  • Master student

Background

Scenexus develops Urban Strategy, an advanced digital twin platform used by cities and regions worldwide for large‑scale scenario analysis in mobility, environment, energy, and climate.

We are developing a next-generation interface where users can interact with and control a complex digital twin using natural language and voice, powered by a locally deployed Large Language Model (LLM) connected to the platform’s Model Control Plane (MCP).

Unlike cloud-based assistants, this AI agent runs in a secure, shielded environment, close to the simulation engine, enabling trusted, explainable, and policy‑aware decision support for public-sector use cases.

The first prototype exists. We are now looking for a Master student to develop this into a robust proof of concept.

Your assignment

You will work on:

  • Designing an LLM-based agent architecture that can interpret user prompts or voice input.
  • Translate them into safe, structured commands for the Urban Strategy MCP
  • Implementing guardrails and constraints to ensure:
    • Simulation integrity
    • Bounded agent behavior (no uncontrolled actions)
  • Deploying and testing the agent in a secure / on‑prem or edge-like environment
  • Demonstrating the concept with realistic urban planning scenarios

 

Depending on your interests, the thesis can focus on questions such as:

  • How do we design safe interaction layers between natural language input and complex simulation systems?
  • What human–AI interaction patterns work best for controlling large‑scale digital twins?

 

Required Skills & Qualifications

  • Master student in AI, Computer Science, Data Science, Robotics or related.
  • Large Language Models, Agent-based systems or AI control loops, Python, APIs
  • Interest in public-sector, smart cities or digital twins is preferable

 

What we offer

  • A cutting-edge AI × digital twin research problem
  • Real-world deployment context with cities and public authorities
  • Close supervision by Scenexus’ technical leadership
  • A lively and internationally focused start-up working environment

 

Interested?

Send your CV and motivation to Walter Lohman, CTO of Scenexus at walter.lohman@scenexus.com