About

Earth Virtual Expert (EVE) is one of Φ-lab's activities in the realm of foundation models. The objective of EVE is to create a Large Language Model AI expert from an open-source LLM, by fine-tuning it on specific and crafted documents related to Earth observation and Earth science.

Earth Virtual Expert (EVE)

EVE will consist of a web platform with a user-friendly graphical interface, complemented by an application programming interface (API) that will serve as the gateway to the trained Large Language Model (LLM). This initiative is part of the European Space Agency's (ESA) broader efforts in Artificial Intelligence for Earth Observation (AI4EO). Φ-lab’s long-term vision for its work in foundation models and LLMs is to pave the way for an Earth Observation (EO) digital assistant capable of responding to information and knowledge queries in natural language, delivering reliable, validated content.
‍
The EVE project is being developed by Pi School, with support from Imperative Space.
Outline of a human head with a brain-like circuit design, symbolizing artificial intelligence and cognitive technology

Partners

‍

‍

Pi School was founded in 2017 by Translated, a global leader in language technology solutions, to address the shortage of skilled AI professionals. Its flagship programme, The Pi School of AI, helps organisations integrate AI into business operations by selecting the world's top engineers and scientists and having them apply their academic knowledge to solving real-world challenges. This is done through an intensive eight-week programme that equips top scientists and engineers with advanced AI skills, focusing on creating MVP prototypes tailored to business needs. Since its inception, Pi School has trained over 230 AI experts and developed 100 MVPs for clients, including Amazon, Meta, ESA and the European Commission.

Pi School is also involved in European Open Science NLP projects alongside Earth Virtual Expert (EVE):

  • Meetween: a multimodal foundation model for speech, focused on videoconferencing use cases funded by the European Commission.
  • Trallie - Transfer learning for information extraction: an LLM solution that converts natural language asset descriptions into structured fields for faceted search, established by the NGI-Search consortium.
  • Videoloc: a UX-based reference-less preference metric for automatic subtitle generation. Funded by Fondazione Bruno Kessler.

Imperative Space is a multi-disciplinary media, education, communications, technology and innovation company working across the space sector. Imperative Space's mission is to support the growth and impact of the sector, especially where it is helping to address global challenges, sustainable development and the furthering of human knowledge.

Imperative Space has worked on projects in Earth Observation, Satcoms, Exploration, Human Spaceflight and Astronomy. Their work ranges across film and multi-media projects, large-scale education and training initiatives, online courses, data visualisation, apps and data platforms, interactive hardware, technical consultancy, strategic communications, design and research.

Links
Follow us
Contact Us
 © EVE 2026
PRIVACY Policy
 AND
Cookie NOTICE