Luma Tech Stack

Multimodal AI for video generation and creative workflows

Software Development San Francisco Bay Area, CA 51–200 employees Privately Held

Luma builds multimodal foundation models (video, 3D, generative media) with a product-first deployment strategy: Dream Machine ships generative video to creators; parallel work on agent stacks and AI-accelerated production workflows suggests a shift from single-modality generation toward agentic systems that operate across domains. The hiring mix is heavily research-forward (17 dedicated research roles against 42 engineering), yet sales and product teams are growing—indicating transition from research-output to commercialization, with noted friction around moving research into production systems and scaling inference.

Tech Stack 96 technologies

Core StackHubSpot PyTorch TensorFlow Figma Adobe Premiere Pro Python Playwright Cypress Selenium GitHub Actions Postman JavaScript TypeScript React Next.js Docker Kubernetes Linux Apollo GAN Outreach JAX pytest LoRA ControlNet ComfyUI CUDA NCCL MPI Triton+61 more

AdoptingRAG GitOps ComfyUI

What Luma Is Building

◆Challenges

Expanding into ksa
Building new function from scratch
Transition research into production systems
Scaling data infrastructure
No existing playbook
No mature pipeline
Defining new category
Scaling multimodal ai models
Accelerating model iteration and evaluation
Scaling complex reasoning engines

▲Active Projects

Dream machine
Develop ksa sales engine
Ai-accelerated production workflows
Project halo
Forward deployed creative playbook
Agent stack
Scaling training & inference
Multimodal foundation models
Large-scale generative ai models
New evaluation frameworks and benchmarking approaches

Hiring Activity

Accelerating120 roles · 70 in 30d

Department

Engineering

Research

Sales

Product

Design

Data

Ops

Seniority

Senior

Mid

Lead

Director

Intern

Junior

Manager

Staff

Notable leadership hires: Delivery Lead, Partner Marketing Lead

Company intelligence

Find more companies like Luma by tech stack, pain points and active projects

Get started free

About Luma

Luma AI develops multimodal foundation models and generative media products targeting creators and production teams. Dream Machine, the flagship product, enables fast visual generation from text and images. The tech stack spans PyTorch, JAX, and TensorFlow for model development; Docker, Kubernetes, and CUDA/NCCL for distributed training and inference; and Next.js/React for product surfaces. Active projects include scaling multimodal training, agent stack development, and production workflow automation. The company is expanding hiring across Peru, United States, United Kingdom, Saudi Arabia, Germany, and Kyrgyzstan, with most roles concentrated in senior and mid-level engineering and research.

HeadquartersSan Francisco Bay Area, CA

Company Size51–200 employees

Hiring MarketsPeru, United States, United Kingdom, Saudi Arabia, Germany, Kyrgyzstan

Frequently Asked Questions

What is Luma AI building?

Luma develops multimodal AI models for video, 3D, and generative media. Dream Machine is their primary product—a text-to-video and image-to-video generator for creators. Active projects include agent stacks, AI-accelerated production workflows, and scaling multimodal foundation models.

What tech stack does Luma use?

Core ML: PyTorch, JAX, TensorFlow, CUDA, NCCL. Infrastructure: Docker, Kubernetes, Linux. Web: Next.js, React, JavaScript, TypeScript. Testing: Pytest, Playwright, Cypress, Selenium. Adopting: RAG, GitOps, ComfyUI.