Remote AI Research Engineer: Agentic Multimodal RL

Jobgether; United Kingdom; Remote; GBP 90,000 - 130,000; Full time

Benefits offered by this job

Remote work

Publish opportunities

Global collaboration

Professional development

Discretionary recognition

Job summary

Jobgether is seeking an AI Research Engineer (Agentic Reinforcement Learning) based in the United Kingdom. The role focuses on advancing reinforcement learning for multimodal AI systems and building scalable training infrastructures.

You will explore diffusion-based generative models, autoregressive architectures, and unified multimodal frameworks, delivering research that pushes model performance, efficiency, and generalization across modalities.

Qualifications

• Master's degree in CS or related field; PhD preferred.
• Strong AI research track record with publications at top conferences.
• Experience leading large-scale RL experiments in multimodal/vision-centric systems.
• Ability to improve domain-specific decision-making with RL.
• Deep understanding of RL algorithms for vision and multimodal learning.
• Knowledge of policy stability, exploration, sample efficiency in high-dimensional modalities.
• Proficiency in PyTorch and modern deep learning frameworks.
• Hands-on experience building end-to-end RL pipelines (simulation to deployment).
• Ability to address RL challenges like sample inefficiency and training instability.
• Experience designing robust evaluation frameworks and empirical improvements.
• Publication record at ICML, NeurIPS, ICLR, CVPR, ICCV, ECCV.
• Strong analytical thinking and scientific problem-solving.

Responsibilities

• Research and develop RL algorithms for multimodal models including diffusion-based and autoregressive approaches.
• Design scalable RL infrastructure for distributed training across complex multimodal systems.
• Develop reward modeling to improve training stability and alignment.
• Create, curate, and maintain multimodal simulation environments and datasets.
• Design and implement rigorous evaluation protocols against baselines.
• Analyze and optimize policy performance across modalities.
• Investigate new RL paradigms to improve learning efficiency.
• Improve exploration, decision-making, and adaptive behavior for state-of-the-art performance.
• Publish and present findings at leading conferences.

Skills

AI research

Reinforcement learning

Multimodal learning

PyTorch

Vision systems

Scientific publications

Education

Master's in CS or related field

PhD preferred

Tools

PyTorch

DL frameworks

Job description

Jobgether is seeking an AI Research Engineer (Agentic Reinforcement Learning) based in the United Kingdom. The role focuses on advancing reinforcement learning for multimodal AI systems and building scalable training infrastructures.

You will explore diffusion-based generative models, autoregressive architectures, and unified multimodal frameworks, delivering research that pushes model performance, efficiency, and generalization across modalities.