Co-op Researcher – LLM Post-Training & Agentic Coding

Founded in 2012, the Noah’s Ark lab has evolved into a prominent research organization with notable achievements in academia and industry. The lab’s mission focuses on advancing artificial intelligence and related fields to benefit the company and society. Driven by impactful, long-term projects, the aim is to enhance state-of-the-art research while integrating innovations into the company's products and services, including Next Generation Foundation Model, Agentic Models, Physical AI and Reinforcement Learning.

• Conduct cutting-edge research on LLM post-training, reinforcement learning, reasoning, agentic coding, and self-improving agents, with the opportunity to take technically challenging ideas from initial hypothesis to rigorous experimental validation.

• Develop new approaches for agent training, task and data synthesis, environment generation, scalable learning from interaction, and agent evaluation, with particular interest in terminal, repository-level, and software-engineering environments.

• Build and evaluate intelligent agents capable of reasoning, tool use, code generation and modification, and solving complex real-world tasks, while studying their capabilities, failure modes, and generalization behavior.

• Own research problems end-to-end, including literature review, problem formulation, experiment design, implementation, large-scale evaluation, analysis, and communicating high-quality research findings.

• Build reliable and scalable infrastructure for LLM training, inference, agent rollout, evaluation, and synthetic data generation, supporting rapid experimentation across large models and datasets.

• Improve the efficiency and reliability of GPU-based post-training and inference workflows, including distributed execution, resource utilization, debugging, automation, and day-to-day research compute operations.

• Work closely with researchers and engineering teams across Huawei to rapidly prototype new ideas, reproduce and extend state-of-the-art methods, and translate promising research advances into practical LLM systems and downstream projects.

• Contribute in a fast-moving, high-ownership research environment where a focused Co-op term can lead to meaningful research, engineering, open-source, or publication-quality contributions.

• Currently enrolled in a senior Bachelor’s, Master’s, or PhD program in Computer Science, Machine Learning, Artificial Intelligence, or a related technical field, with strong fundamentals in machine learning and modern deep learning.

• Strong understanding of large language models and modern post-training techniques, including Transformer-based models, supervised fine-tuning, reinforcement learning, preference optimization, distillation, reasoning, or synthetic-data-driven training.

• Strong programming and software-engineering skills, particularly in Python and PyTorch, with the ability to independently implement, debug, profile, and iterate on complex ML or agent systems.

• Hands-on experience in one or more relevant areas, such as LLM post-training, coding or tool-using agents, reinforcement learning, large-scale model training/inference, distributed systems, or ML infrastructure.

• Demonstrated ability to take an open-ended and ambiguous technical problem, identify the important questions, design effective experiments or systems, and independently drive the work toward measurable results.

• Strong analytical and experimental judgment, including the ability to design controlled experiments, evaluate models rigorously, diagnose unexpected behaviors, and distinguish genuine improvements from experimental noise.

• Highly self-motivated, intellectually curious, and execution-oriented, with strong communication and collaboration skills and the ability to operate effectively in a rapid research iteration cycle with substantial ownership.

• Strong assets include top-tier AI/ML publications or substantial research experience, impactful open-source or agent/LLM projects, and hands-on experience with frameworks and systems such as vLLM, SGLang, FSDP, DeepSpeed, Hugging Face, Docker, Linux, Git, or multi-GPU compute environments; we welcome candidates who are especially strong in either research or systems engineering, and particularly value those who can bridge both.

Huawei Canada is committed to a fair, inclusive, and accessible recruitment process. If you require accommodation during any stage of the hiring process, please let us know and we will work with you to meet your needs.

All applications for this position are reviewed directly by our hiring team, we do not use artificial intelligence tools to screen or select candidates.