L
Luma Tech Stack
Multimodal AI models for video generation and creative workflows
Luma builds foundational multimodal models trained on distributed GPU/TPU infrastructure (PyTorch, CUDA, Ray, Slurm) and ships them in consumer products like Dream Machine. The hiring mix is heavily research-forward (9 research roles among 66 total), with early-stage sales and GTM infrastructure (6 posted sales/marketing/partner roles in the last 30 days), suggesting a company scaling from model development toward revenue — pain points around 'no existing playbook' and 'no mature pipeline' confirm they're still establishing go-to-market and customer success from first principles.
Tech Stack 96 technologies
Core StackPyTorch AWS Python Go Kubernetes TensorFlow Ashby Greenhouse Lever Terraform Ansible Prometheus Grafana Canvas CUDA NCCL MPI OCI NVIDIA AMD InfiniBand RDMA Bash GPU TPU Ray Slurm JAX GCP ComfyUI+62 more
What Luma Is Building
◆Challenges
- No existing playbook
- No mature pipeline
- Scaling research into production
- Expanding into entertainment vertical
- Building new function from scratch
- No legacy process
- Scaling customer success function
- Expanding platform usage
- Building customer success processes
- Building sales pipeline
▲Active Projects
- Develop german sales engine
- Early revenue engine
- Project halo compute supercluster
- Forward deployed creative playbook
- Control interface of foundation models
- Early go-to-market motion
- Large-scale multimodal model training
- Shape platform evolution
- Ai-accelerated production workflows
- Customer success processes
Hiring Activity
Steady65 roles · 20 in 30d
Department
Seniority
Notable leadership hires: Partner Marketing Lead
Company intelligence
