About the role
You'll work alongside researchers and engineers building state-of-the-art foundation models, scalable AI infrastructure, and intelligent data systems that power the next era of scientific innovation. This is a rare opportunity to contribute to technology that sits at the intersection of Deep Learning, LLMs, Distributed Systems, and Scientific Computing.
Responsibilities:
- Designing and optimising Foundation Model training pipelines.
- Building and fine-tuning Transformer and LLM architectures.
- Developing intelligent data curation, preprocessing, and tokenisation pipelines.
- Implementing evaluation frameworks, benchmarking systems, and model quality metrics.
- Working on RLHF, alignment, and post-training optimisation workflows.
- Building multilingual and scientific datasets for next-generation AI models.
- Collaborating with infrastructure teams on distributed GPU/TPU training systems.
- Improving scalability, performance, and reliability of production AI platforms.
Requirements:
- Strong Python programming skills.
- Solid understanding of Machine Learning and Deep Learning fundamentals.
- Hands-on experience with PyTorch (or equivalent frameworks).
- Good understanding of Transformers, LLMs, and Foundation Models.
- Knowledge of NLP preprocessing and data pipeline design.
- Strong debugging, analytical, and problem-solving abilities.
- Comfortable working in Linux-based development environments.
Bonus Points If You Have Experience With:
- DeepSpeed, FSDP, or Distributed Training.
- RLHF, DPO, Alignment techniques.
- CUDA programming or GPU optimisation.
- PySpark and large-scale data engineering.
- Scientific ML or numerical computing.
- Dataset curation, evaluation, and benchmarking.
- HPC environments and GPU clusters.