LLM scientist

About the job

Pi School is developing a text-based Large Language Model for Earth Observation and Earth Science on a collaborative research project for the European Space Agency. We are looking for an NLP specialist capable of pursuing independent research experiments and help develop the LLM and LLM-based agent of this project. The researcher should work on-site at our headquarters in Rome.

Go to the application form.

Project specifics

Develop an open-source LLM Earth Observation (EO) Digital Assistant for the European Space Agency (ESA) – EVE – Earth Virtual Expert: X and LinkedIn.
15-month, €1 million project.
Open science project with open source artefacts, resulting in LLM assistant to be publicly accessible on ESA Φ-lab EO innovation lab.
Development and research on the entire life cycle of the project: Data Collection, Continued Pre-Training, Instruction Fine-Tuning, Retrieval, Augmented Generation.
Focus on hallucination reduction and empirical knowledge.
Research state-of-the-art techniques and develop innovative solutions.

At the end of this project, the candidate could continue working on similar open-science NLP projects that Pi School is working on in parallel at the moment, as well as future ones:

Meetween – multimodal foundation models for speech, use-cases for videoconferencing.
Trallie – LLM solution to turn natural language asset descriptions into structured fields for faceted search.
Videoloc – UX-based reference-less preference metric for automatic subtitle generation.

What You’ll Do

You will be part of Pi School’s AI Research team full-time, focusing on various technologies including Large Language Models, as well as Machine and Deep Learning. In this role, you will:

Design novel research experiments, and be able to craft prototypes or early results to inform the AI team’s decision-making on the best options for the development of the mode.
Monitor and benchmark the state of the art on LLMs and related NLP areas.
Familiarity with Docker and Linux, including running GPU experiments.
Contribute to the development of our project: work with data, compute and algorithms; design of deep learning neural architectures; fine-tune, customise and improve pre-trained Large Language Models; execute experiments on high computing platforms (GPU on HPC) and conduct evaluations; improve model reliability and mitigate biases and hallucinations; adapt the project’s pace to the rapid scientific development in our field.
Organise publications and efforts related to open-sourcing.

Ideal qualifications

Recent publications on LLMs and NLP (bonus if at top-tier ML/AI conferences such as NeurIPS, ICML, ACL, or EMNLP), as well as open-source contributions.
Hands-on experience with: LLMs like Mistral, Llama, or Gemma families; end-to-end orchestration tools like Langchain, or NVIDIA NeMo; fine-tuning libraries like Axolotl, or Unsloth
computing platforms like Runpod, DeepInfra, or LambdaLabs; RAG databases like Qdrant, FAISS, or Pinecone
Excellent programming skills in Python, ML/DL frameworks, especially Pytorch.
Familiarity with Docker and Linux, including running GPU experiments.
Excellent proficiency in English.

About Pi School

Pi School is on a mission to empower businesses and professionals to leverage AI’s transformative capabilities. We believe technology should be an enabler, not a barrier, and it should empower individuals and businesses to grow, innovate, and stay connected. We strive for a world where AI and human potential coexist harmoniously, reshaping the way we work, learn, and prosper.

Born in 2017 from the expertise of Translated, an industry pioneer with over two decades of leadership in AI + human translation, Pi School has helped over 100 companies integrate AI solutions to improve their operations, enhance decision-making, and achieve sustainable growth.

Pi School remains at the forefront of the ever-evolving AI landscape and collaborates with prominent research centres and organisations worldwide, such as the European Space Agency (ESA) and the European Commission, to harness AI for both cutting-edge research and practical solutions, driving progress in science and beyond. Our approach is founded on purposeful, impactful work, leveraging technology as a force for good.

Our office

Pi School is hosted at Pi Campus, a working environment immersed in nature. Six luxury villas in Rome (Italy) have been converted into functional offices to foster talent growth. Pi Campus is also a venture firm created by Translated to reinvest part of its profits into promising AI startups.

Benefits and Perks
Our working environment is both relaxed and intense. We are passionate about our mission, and our work is highly regarded in our industry.

Fresh fruit and snacks.
Access to the pool and Turkish bath.
Sports activities: gym and pilates.
Free time in between your tasks to work on your personal projects.
Access to the Campus parties and gatherings.

Learn more about our company: https://picampus-school.com/

Diversity

At Pi School, we proudly embrace and celebrate each individual’s unique qualities to our team, regardless of race, sexual orientation, gender identity, or any other differences. We recognise that these diverse perspectives empower us to overcome challenges, foster innovation, and drive excellence. As an inclusive and equal-opportunity employer, we are committed to cultivating an environment where everyone feels welcome, valued, and supported to achieve their full potential.