Explore calibration for large language models with the National Research Council in Ottawa. Bring experimental design and scientific writing expertise.
As a Postdoctoral Fellow, you will develop methods that make LLM uncertainty more meaningful and interpretable. The two-year research project examines generated text and intermediate features, addressing reproducibility, embedded knowledge, and trustworthy AI. Research outcomes may target ACL, ICLR, or ICML venues.
Key Responsibilities:
- Consolidate probability distributions into interpretable certainty metrics
- Study reproducibility and embedded knowledge in LLM predictions
- Develop calibration methods for intermediate model features
- Support evaluation and trustworthy decision support applications
- Produce first-author publications and present research findings
Requirements:
- Required PhD in a specified quantitative or computing discipline
- Experience designing speculative AI-related research projects
- Deep learning tools experience for collection, analysis, or simulation
- High-calibre publication experience in design, writing, and presentation
- English language proficiency and Reliability Status eligibility
Use calibration, experimental research, and scientific communication to clarify uncertainty across large language model systems.
📌 Postdoctoral Researcher LLM Calibration (Ottawa)
🏢 NRC
📍 Ottawa