World Models: Multiobjective Policy Composer

Italian Ministry of Education, University and Research; Italia; In loco; EUR 60.000 - 75.000; Tempo pieno

Descrizione del lavoro

Unimore in Italy invites applications for a recognised researcher position focused on research and implementation of composable policies for multi-objective optimization in World Models (model-based reinforcement learning). The role aims to combine policies learned on separate objectives at inference time without retraining.

The successful candidate will engage in selective procedures based on qualifications and interview; country/ies of residence: OTHER.

Competenze

• Experience in research on reinforcement learning and policy optimization.
• Familiarity with model-based approaches and world models is desirable.

Mansioni

• Research and implement composable policies for multi-objective optimization in World Models.
• Explore objective trade-offs and task arithmetic in weight space for inference-time deployment.

Conoscenze

World Models

Composable policies

Multi-objective optimization

Reinforcement learning

Model-based RL

Descrizione del lavoro

Unimore in Italy invites applications for a recognised researcher position focused on research and implementation of composable policies for multi-objective optimization in World Models (model-based reinforcement learning). The role aims to combine policies learned on separate objectives at inference time without retraining.

The successful candidate will engage in selective procedures based on qualifications and interview; country/ies of residence: OTHER.