AI Research Engineer (Multi-Modal Reinforcement Learning)
New
United StatesFull-TimeMiddle
Salary not disclosed
Apply NowOpens the employer's application page
Job Details
- Required Skills
- Machine LearningPyTorchDeep LearningNLPComputer Vision
Requirements
- Master’s degree in Computer Science or related field required; PhD preferred in ML, CV, NLP, or AI-related disciplines.
- Strong publication record in top-tier AI conferences (NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV).
- Proven experience in large-scale reinforcement learning experiments, particularly in multi-modal or vision-centric systems.
- Deep understanding of reinforcement learning theory, optimization, and policy learning in high-dimensional environments.
- Strong hands-on experience with PyTorch and deep learning frameworks for multimodal AI systems.
- Experience building end-to-end RL pipelines including simulation, training, evaluation, and deployment.
- Ability to address core RL challenges such as sample efficiency, exploration-exploitation trade-offs, and training stability.
- Strong analytical and problem-solving skills with a research-driven, experimental mindset.
Responsibilities
- Conduct research on reinforcement learning methods for multi-modal systems, including diffusion-based and autoregressive model approaches.
- Design and build scalable RL infrastructure supporting distributed training and evaluation across complex multi-modal environments.
- Develop reward modeling strategies to improve alignment, training stability, and mitigate failure modes such as reward hacking.
- Create and curate simulation environments and datasets for training, benchmarking, and validating multi-modal RL models.
- Design and execute evaluation protocols to measure performance improvements and ensure reproducibility across experiments.
- Analyze model behavior across modalities, identifying bottlenecks in optimization, exploration, and cross-modal alignment.
- Explore and develop next-generation RL paradigms to enhance learning from environment feedback and improve SOTA performance.
- Publish research in leading AI conferences such as NeurIPS, ICML, ICLR, CVPR, and related venues.
View Full Description & ApplyYou'll be redirected to the employer's site
