About the role#
The AI, Learning and Intelligent Systems group is seeking a graduate researcher to focus on LLM reliability and uncertainty. You will investigate how to quantify uncertainty in AI science assistants, specifically during multi-turn scientific dialogues. The goal is to identify when scientific tasks are underspecified or ill-posed, preventing models from making unverified assumptions.
What you'll need#
- Minimum 3.0 cumulative GPA.
- Familiarity with large language models, including multi-turn or agentic systems.
- Experience developing or evaluating machine learning models.
- Knowledge of probabilistic machine learning or uncertainty quantification.
- Hands-on experience with open-weight LLMs and deep learning frameworks.
- Ability to run Python code on HPC or multi-GPU systems.
- Strong software engineering and debugging skills.
- Currently enrolled as a full-time graduate student or have earned a degree within the past 12 months.
What you'll do#
- Research and evaluate methods for uncertainty quantification and hallucination detection in scientific workflows.
- Develop probing methods to predict when a scientific task is incomplete or inconsistent.
- Build and instrument evaluation pipelines to analyze model internal states on HPC systems.
- Conduct experiments and analyze model behavior across various scientific domains.
- Contribute to technical documentation, research reports, and project presentations.
Location & details#
- Location: Golden, Colorado (Remote modality available).
- Position type: Full-time, paid internship.
- Term: Rolling.
- Note: All candidates must be able to obtain and maintain a federal Personal Identity Verification card and pass a pre-employment drug test.


