Staff Software Engineer, Product

About Luma

A new class of intelligence is emerging, systems that understand and generate the world across video, images, audio, and language. Building multimodal AGI is not just a research problem. It's a full-stack engineering problem, spanning training systems, inference, product architecture, and the tight feedback loops between them.

When a new technological wave begins, the highest leverage place to be is at the foundation. Not at a startup wrapping someone else's API. At the company building the models themselves.

The frontier is not incremental improvement. These models are replacing entire categories of software and enabling entirely new ones.

At Luma, we are:

• A leading multimodal AI research lab, having built one of the world's strongest video generation models (Ray-3.14).
• Pushing beyond video toward the next generation of multimodal general intelligence models.
• Operating at a scale few companies can match, with the compute and resources to support frontier research ($900M Series C).
• Focused on the creative domain, where multimodal systems can have immediate real-world impact.
• Shipping tightly integrated products that turn research breakthroughs into tools creators actually use.

We're still early. The playbook is not written. A single exceptional engineer can reshape how the company operates.

Where You Come In

We are looking for senior engineers who understand that the game has changed. The era of humans writing code is over. AI makes implementation cheap. Decision-making is now the constraint:

• What to build.
• What not to build.
• Where to draw boundaries.
• What problems deserve structure, and which should stay flexible.

With AI tools, weak decisions compound faster. Speed amplifies both clarity and mistakes. We need technical leaders with exceptional judgment, engineers who can direct AI to move quickly while making disciplined choices about scope, structure, and tradeoffs.

This role is not about writing more code. It's about deciding where code should exist at all. You will spend your time on the decisions that shape velocity:

• Translating ambiguous goals into clear technical direction.
• Choosing when to invest in structure and when to stay lightweight.
• Preventing unnecessary complexity from entering the system.
• Aligning product intent with technical reality.
• Teaching others how to use AI tools without overbuilding.

AI can generate solutions to any prompt. It cannot decide which problems are worth solving. That responsibility is yours.

The ideal candidate has deep technical judgment (system design, architecture, security) and strong product intuition (user needs, prioritization, UX). What's non-negotiable: you must be world-class at directing AI agents to achieve real product and engineering outcomes, and at guiding technical direction for a team operating at high speed.

What You'll Do

• Direct AI coding agents to build production systems quickly, while maintaining discipline around system design, ownership, and real-world behavior.
• Decide what to build and what not to build as implementation costs drop.
• Set technical direction across engineering, including system boundaries, data models, ownership, and operational expectations.
• Co-own product direction in partnership with design and research, defining both what we build and how we build it.
• Make and stand behind hard tradeoffs between speed, reliability, complexity, and extensibility.
• Shape architecture early , before decisions harden into constraints.
• Own critical outcomes end-to-end , from problem framing to deployment and iteration.

Raise the Bar

• Set the standard for technical depth, judgment, and production ownership.
• Prevent avoidable complexity from entering the system as iteration speed increases.
• Mentor engineers on how to use AI tools effectively without lowering engineering rigor.
• Identify and eliminate recurring classes of design mistakes before they scale.

Multiply the Organization

• Design systems and workflows that allow small teams to ship with outsized impact.
• Attract exceptional engineers and help them grow into high-judgment technical leaders.
• Increase team velocity without increasing operational and conceptual overhead.

Who You Are

Required:

• You have 8+ years of experience building and shipping production software.
• You have led multiple non-trivial projects, owning technical direction and execution end-to-end.
• You have made high-stakes technical decisions that materially affected a system or product.
• You are fluent with AI coding tools and have used them to ship real functionality in production or substantial side projects.

You possess either:

• Strong technical depth: System design, distributed systems, security, infrastructure.
• Strong product intuition: User empathy, prioritization, taste in UX.
• (Ideally, both)

What Sets You Apart

You've been personally responsible for at least one clearly measurable, real-world success. For example:

• You led a product that reached 100k+ users or 1M+ monthly active users.
• You designed and led a production system serving millions of requests per day.
• You created or led an open-source project with 10k+ stars or meaningful external adoption.

Why This Role Exists

Most engineering roles optimize mature products. This one defines how software gets built inside a frontier multimodal AI lab.

The decisions you make here will shape:

• How research turns into products.
• How those products scale.
• How fast we can move without losing rigor.
• How engineering operates in an AI-native company.

If you want to help set the standard for technical leadership in this new era of AI, we should talk.

Compensation

The base pay range for this role is $230,000 - $360,000 per year.

About Luma

Luma's mission is to build unified general intelligence that can generate, understand, and operate in the physical world.

We believe that multimodality is critical for intelligence. To go beyond language models and build more aware, capable and useful systems, the next step function change will come from vision. So, we are working on training and scaling up multimodal foundation models for systems that can see and understand, show and explain, and eventually interact with our world to effect change.