Research Scientist - Reinforcement Learning

Develop novel research toward massive scale self-play for foundation model training, agentic tasks, and imbuing models with the capability to proactively learn from its environment. 

Initiate and pursue novel reinforcement learning algorithmic approaches to define and drive emergent capabilities in Foundation Models. 

Full-stack engineering from data curation, model architecture and algorithm design, to final production of models for end-users using high quality (documented, tested, maintainable) code. 

Contribute to technical reports and research publications. 

Represent MBZUAI at industry conferences and events, showcasing the institution’s technology and deep learning capabilities and establishing MBZUAI as a global leader in AI research and innovation. 

Proactively engage with the open-source community. 

Contribute to large-scale reinforcement learning training and inference frameworks. 

Facilitate internal and external collaboration 

Demonstrated ability to independently identify limitations of current practice (internal and external), formulate and enact solution strategies for improvement. 

Proactive mindset with the ability to identify impactful research questions and execute on them with minimal supervision. 

Strong Python development skills with a focus on research-grade code and scalable data pipelines. 

Practical experience implementing complex mathematical concepts into reliable, well-documented code. 

Experience applying novel RL algorithms to practical applications. 

Strong experience contributing to academic and/or open-source research through publication, GitHub contributions, or professional presentations. 

Strong communication and collaboration skills for effective cross-functional work. 

Strong systems and engineering expertise in deep learning frameworks such as PyTorch, Jax, etc. 

Experience in large-scale model training (LLMs or Diffusion Models) on large clusters. 

Familiarity with current RL+LLM training libraries 

Experience training policies in self-play, possibly demonstrated by publication, blog post, public code. 

Experience working with Diffusion Models in RL, possibly demonstrated by publication, blog post, public code. 

Strong publication record in leading AI and RL venues (e.g.ICLR, ICML, NeurIPS, RLC, JMLR, TMLR) 

Familiarity with performance constraints in production environments and the trade-offs in model design and execution. 

Prior contributions to open-source ML research or data tools. 

Demonstrated ability to solve complex system-level challenges and debug failures across training/inference stack (e.g. memory issues, deadlocks, I/O bottlenecks, multi-node communication failures).