Research, identify and evaluate state-of-the-art LLMs for effective, correct, and grounded edge-based mission planning.
Your role
The key responsibilities of this role include conducting a review on existing LLMs and their applications in edge computing. This involves identifying the strengths and limitations of current LLMs in the context of mission planning and reviewing case studies and existing implementations of LLMs in similar domains. The selected LLMs will be evaluated and documented based on criteria developed for assessing their effectiveness, correctness, and grounding in the application of mission planning on the edge.
Background
The research field and applications of Large Language Models has literally exploded during the last years. This has led to models that are getting more and more capable whilst their size and hardware requirements continue to grow smaller. While more and more capable, at the time of writing, most state-of-the-art open-source models, still does not have the performance that is required to add true value in the application of edge-based mission planning. E.g., they lack in inference performance, quality of the results, or they are still too resource-heavy for effective execution on an edge device.
Description of the master thesis
This work aims to study the field of open-source LLM’s and too narrow down on models that are truly capable in the application of edge-based mission planning. This work will include identification and evaluation of identified models and to assess their performance based on measures relevant for mission planning on the edge. The identified models will be evaluated based on criteria developed for assessing their effectiveness, correctness, and grounding in in the mission planning domain. Prototypes or proof-of-concept systems will be developed to demonstrate the application of LLMs in edge-based mission planning, and their performance and scalability will be evaluated.
This work will include the following work packages:
1. Identification of state-of-the-art in open-source LLM’s relevant for edge-based mission planning.
2. Identification of relevant measures and tests for validation of identified models.
3. Demonstration and documentation of the results
Your profile
This Master Thesis is suitable for one student. You are at the end of your master studies in computer science, software engineering, or applied mathematics. A specific interest in AI and planning applications is merited.
This position requires that you pass a security vetting based on the current regulations around/of security protection. For positions requiring security clearance additional obligations on citizenship may apply.
We provide the support and guidance you need to translate your theoretical knowledge into practical solutions. Join us and become a driving force behind Saab's technological advancements!
What you will be a part of
The Aeronautics business area is an innovative supplier of world-class aircraft systems, advanced aerostructures, and a wide range of support solutions within civil and military aviation. The business area researches, develops, and produces military aviation systems. We are building for the future through research and studies into innovative flight systems and the further development of our products. Read more about us here, https://www.saab.com/about
Contact information
Karina Wandt, Manager
073-418 55 84
karina.wandt@saabgroup.com
Ella Olsson, Master Thesis Supervisor
ella.olsson@saabgroup.com