Copy of Research Scientist / Engineer – Training Infrastructure
Luma AI · London, UK · Remote, UK
CHECKED 23H AGO
Applied AIremote
At a glance
Highlights
- Build systems training large-scale multimodal models on thousands of GPUs.
- Own training stability and efficiency for massive-scale model runs.
- Build monitoring and debugging tools for large-scale training runs.
Ideal candidate
A systems-minded engineer who thrives on large-scale reliability challenges and has solved real foundation-model training problems at scale. You care about efficiency, stability, and enabling researchers to innovate.
Requirements
- Extensive distributed PyTorch training and parallelisms in foundation-model training
- Deep understanding of GPU clusters, networking, and storage systems
- Familiarity with communication libraries (NCCL, MPI) and distributed-system optimization
- Experience with FSDP and multi-node training
Lab profile
About Luma AI
Builds generative AI models for photorealistic video and 3D creation.
What they build
- Dream Machine video generator
- Ray2 video model
- Photon image model
Known for
- Photorealistic video synthesis
- 3D world and object generation
- Large-scale training infrastructure
- High-performance inference optimization
Key facts
- Founded in 2021
- Headquartered in San Francisco, California
- Backed by Andreessen Horowitz