Master’s Thesis: Demystifying LLM Reasoning Traces

AI; Göteborgs kommun; Hybrid; SEK 201,000 - 268,000; Full time

Benefits offered by this job

Hybrid working mode

Access to substantial compute

Job summary

AI Sweden, Sweden’s national center for AI, invites a Master’s thesis student to join in collaboration with Stef SM Lab at Chalmers. The role focuses on adversarial robustness and honesty in reasoning traces, with work arranged as a thesis project at Lindholmen Science Park, Gothenburg, in hybrid mode.

You will design experiments, run open-weight reasoning models, and contribute to frontier AI safety research with access to substantial compute. Start date January 2027, deadline 2026-12-27.

Qualifications

• Ongoing Master’s studies in Computer Science, Data Science, Engineering Physics, Complex Adaptive Systems, Machine Learning, or similar.
• Comfortable with Python and deep learning frameworks; open to surprising experimental results.
• Prior experience with language models and reinforcement learning is a plus, but genuine curiosity and interest in AI safety and security matter most.

Responsibilities

• Design controlled experiments to understand the reasoning phenomenon.
• Run experiments on open-weight reasoning models and measure outcomes.
• Collaborate with AI Sweden researchers and leverage substantial compute resources.

Skills

Python

Deep learning

AI safety interest

Experiment design

Education

Master’s student in Computer Science/Data Science/ML

Tools

PyTorch/TensorFlow

Job description

AI Sweden, Sweden’s national center for AI, invites a Master’s thesis student to join in collaboration with Stef SM Lab at Chalmers. The role focuses on adversarial robustness and honesty in reasoning traces, with work arranged as a thesis project at Lindholmen Science Park, Gothenburg, in hybrid mode.

You will design experiments, run open-weight reasoning models, and contribute to frontier AI safety research with access to substantial compute. Start date January 2027, deadline 2026-12-27.