Thesis Work 15hp-30hp - AI agents for software development, for single tasks or in cooperation

Thesis project at Scania is an excellent way of making contacts for your future working life. Many of our current employees started their career with a thesis project.

Background:
Are you ready to explore how AI can revolutionize software development beyond just writing code? Large language models (LLMs) have already proven they can generate source code, but what about the rest of the development process? This thesis project challenges you to build AI agents capable of performing all the key tasks in a software project — from project management to testing, and also to make AI agents collaborate, in order to solve bigger more complex SW development tasks.

Target:
The specific tasks we intend to focus on are: project management, resource estimation, reviewing, planning, architecting, requirements writing, programming, and testing. In some of these tasks, an agent might utilize not only LLMs, but also formal methods. The assignments below are intended to, either focus on one such task, or on setting up a whole team of agents performing different tasks. One possible approach for this is the MetaGPT framework (GitHub - geekan/MetaGPT: The Multi-Agent Framework: First AI Software Company, Towards Natural Language Programming).
   The students are expected to combine state-of-the-art approaches form research literature with general state-of-the-art services available on internet. The work should benchmark different alternative solutions, and also benchmark or at least discuss the performance in relation to humans.

 Example of assignments:
- Build a “Project Manager Robot” by using generative AI.
- Set up a team of cooperative AI agents with different roles to develop a bigger more complex software.
- Explore how humans can collaborate and be included in a project team of mostly AI agents.
- Build a time estimation agent, i.e. use generative AI to estimate task time. Benchmark with estimation made by Senior Developers.
- Investigate how LLM based code generation is affected by the type of specifications (e.g. w.r.t. complexity, size, granularity, completeness, level of formality, etc.)

Education:
Type of project: MSc or BSc
Education: computer science, electrical engineering, engineering physics or similar.

Number of students: 2-5
Time: 10-20 weeks, full time 40 hours per week
Start: Jan 2025
Credits: 15hp and 30hp

Contact persons:
Mattias Nyberg, Adj. prof, KTH / Research Manager, Scania,  mattias.nyberg@scania.com

Application:
Enclose CV, personal letter and transcript of grades.

A background check might be conducted for this position. We are conducting interviews continuously and may close the recruitment earlier than the date specified.