Axpo

Development of Large Language Models for safety-related, plant-specific nuclear documentation using Machine Learning techniques

Axpo is driven by a single purpose – to enable a sustainable future through innovative energy solutions. As Switzerland's largest producer of renewable and nuclear energy and an international leader in energy trading and the marketing of solar and wind power, we combine the experience and expertise of more than 6000 employees driven by a passion for innovation, collaboration and impactful change. Through cutting-edge technologies, we innovate to meet the evolving needs of our customers in over 30 countries across Europe, the US and Asia.

Department
Nuclear Safety Department, Leibstadt Nuclear Power Plant, Switzerland

Your responsibilities
The documentation related to licensing and operation of a nuclear power plant operator is extensive. This industrial internship project is part of a program that develops innovative tools in support of the nuclear safety experts for their daily work. The internship shall generate a training database comprising nuclear documentation and develop a program environment comprising available open source large language model (LLM) elements. Following tasks are foreseen for a first 3-month period:

  • Literature survey: conducting a thorough review of related research & development projects (international, as well as in sister companies like Axpo) prior to starting development for the Leibstadt nuclear power plant (KKL)
  • Data base generation: identifying a selection (several hundreds) of relevant safety-related nuclear documents (generic documentation from open sources as well as plant-specific documentation available in the KKL Document Management System). Those documents will be extracted using available scripts and digitalized using a fast and robust OCR tool(s) (to be selected).
  • Program environment development: performing a survey of currently available open source LLM models on the market. The models shall be characterized, evaluated and classified. A proof of concept (PoC) shall be performed for the program environment comprising selected LLMs.
  • Custom train the model on selected data using Python & API like OpenAI API. Define a suitable training architecture.

Your profile

  • Good programming skills in Java, C# or Python.

We offer
You will have the opportunity to work alongside nuclear experts and apply your academic and research background in real industrial applications. In this project you will have the opportunity to work in a hybrid mode (office work in main headquarters in Baden and home-office). Depending on the outcome of the project, the internship could be extended and/or a master thesis topic could be proposed.

Contact

Axpo Power AG offers 18 internship proposals for the Summer 2024.
Please, apply online using the following link by 21.02.2024
In your online application you can specify for which of Axpo’s projects you are interested in.
A selection day will follow at beginning of March at Axpo’s HQ.

Bleiben Sie auf dem Laufenden

Abonnieren Sie unseren Newsletter

Zur Newsletter-Anmeldung

Profitieren Sie als Mitglied

Werden Sie Mitglied im grössten nuklearen Netzwerk der Schweiz!

Vorteile einer Mitgliedschaft