PhD: LLM-Guided SAT/MaxSAT Core Algorithms

University of Lleida; España; On-site; EUR 33,000 - 38,000; Full time

Benefits offered by this job

Fully funded PhD (MSCA)

Enrollment in Doctoral Programme

International research environment

Job summary

The University of Lleida invites applications for a fully funded PhD position in the Logic & Optimization group, under Dr. Carlos Ansótegui.

The successful candidate will research at the intersection of SAT, MaxSAT, proof complexity, and large language models. This three-year MSCA-funded project aims to transform natural-language problem descriptions into formal MaxSAT models, explore core sequences in core-guided MaxSAT, and apply results to interpretable machine learning problems.

Qualifications

• Master's degree in Computer Science, Mathematics, or closely related field by start date.
• Strong background in algorithms, data structures, and computational complexity.
• Solid programming skills in Python and/or C++.
• Familiarity with propositional logic; SAT/MaxSAT exposure is a plus.
• Interest in combining theory with solver implementation and experimentation.

Responsibilities

• Carry out research at the intersection of SAT, MaxSAT, and LLMs.
• Develop problem modelling and encodings with LLM guidance.
• Experiment with solvers and track performance on benchmarks.
• Publish results and contribute to open-source Paramita integration.

Skills

Master's degree

Algorithms

Data structures

Python

C++

Propositional logic

SAT/MaxSAT

Motivation for theory and impl

Education

Master's degree or equivalent

Tools

Paramita framework

Python

C++

Job description

The University of Lleida invites applications for a fully funded PhD position in the Logic & Optimization group, under Dr. Carlos Ansótegui.

The successful candidate will research at the intersection of SAT, MaxSAT, proof complexity, and large language models. This three-year MSCA-funded project aims to transform natural-language problem descriptions into formal MaxSAT models, explore core sequences in core-guided MaxSAT, and apply results to interpretable machine learning problems.