Nscale is looking for AI Engineers to join our core AI team and drive the development of training, fine-tuning, and inference products for our GenAI cloud platform. In this role, you will design and optimise scalable systems for generative AI models, tackle complex performance challenges, and implement cutting-edge AI solutions. This is a unique opportunity to redefine how generative AI is built and consumed globally, working alongside a world-class team renowned for record-setting performance and a strong commitment to open-source innovation.

Responsibilities

• Collaborate with researchers, engineers, and product teams to develop innovative platform features and integrate cutting-edge advancements.
• Architect, build and optimise scalable, high-performance systems for training, fine-tuning, and inference of generative AI models.
• Design and implement advanced methodologies like LoRA, prefix-tuning, and adapter-based approaches for fine-tuning AI models.
• Develop robust, fault-tolerant systems for data ingestion, processing, and model customisation.
• Optimise GPU utilisation and system performance using frameworks like DeepSpeed, Triton Inference Server, TensorRT, and custom CUDA/Rocm kernels.
• Conduct performance testing and resolve bottlenecks across training, fine-tuning, and inference workflows.
• Document and build tooling to ensure successful use of Nscale’s training, fine-tuning, and inference APIs.

Requirements

• 5+ years of experience building and deploying machine learning systems in production environments.
• Proficiency in Python and PyTorch, with a strong understanding of transformer architectures, LLMs, and multimodal generative models.
• Expertise in distributed training frameworks like DeepSpeed or Fully Sharded Data Parallel (FSDP).
• Experience with GPU programming and optimisation (e.g., CUDA, TensorRT, or ROCm).
• Knowledge of fine-tuning methods, such as LoRA, prefix-tuning, and adapter-based techniques, and experience improving model performance.
• Experience with containerised environments and Kubernetes for managing AI workloads, including tools like Kubeflow.
• Familiarity with inference systems such as Triton Inference Server, vLLM, or TGI.
• Demonstrated ability to write performant, well-tested, and production-quality code.
• Passionate about building tools and systems that enhance the AI developer experience.

Preferred

• Experience developing large-scale and high-load production systems.
• Contributions to open-source AI frameworks or inference systems.
• Hands-on experience with advanced inference optimisation techniques, such as KVCache, MoE, adaptive batching, or gradient checkpointing.
• Strong understanding of low-level operating systems concepts, including multi-threading, memory management, and performance tuning.
• Experience developing APIs using OpenAPI 3.0+ specifications.
• Knowledge of efficient training and inference evaluation strategies, with demonstrated success in improving model efficiency.

Core Values

• Relentless Innovation: We constantly push the boundaries of innovation, embracing creative risks to shape the future.
• Ownership and Accountability: Every Nscaler is fully accountable for their work, driving it with excellence and urgency.
• Openness and Transparency: We maintain open communication within our teams and with stakeholders, sharing both successes and challenges.
• Customer-Centric Focus: Our customers are central to our mission, and we are committed to delivering impactful solutions that drive real-world success.
• Sustainability: We are dedicated to considering the long-term environmental and societal impacts of our technologies.
• Full-Speed Collaboration: Collaboration at Nscale is fast, efficient, and respectful.

Equal Opportunities Statement

At Nscale, we are committed to fostering an inclusive, diverse, and equitable workplace. We believe that a variety of perspectives enriches our work environment, and we warmly welcome applications from individuals of all backgrounds, experiences, and perspectives. We strongly encourage applications from people of colour, the LGBTQ+ community, people with disabilities, neurodivergent people, parents, carers, and people from lower socio-economic backgrounds.