Post-Doctoral Research Visit F/M Why Do Protein Structure Predictors Confidently Produce Nonsense, and Can We Anticipate These Failures?

Investigate hallucinations and physical plausibility failures in protein language models. Develop explainability methods connecting model mechanisms and input sequences to predicted structures, using the FLoPS dataset and abstraqt annotations. The project combines machine learning, bioinformatics and biophysics, with possible exploratory work on flow-matching predictors. This renewable, 12-month appointment in Sophia Antipolis is planned from 4 January 2027. Gross monthly pay is EUR 2,788.

A PhD or equivalent and fluent English. The work requires designing reproducible computational analyses and communicating interdisciplinary findings. Restricted-area security authorisation may be required.