Job Description

About Atlas

Atlas is building an AI-native creative platform for professional 3D and game-
production workflows.

Our product combines an agentic canvas, generative models, reusable workflows,
and production integrations to help creators move from ideas and references to
usable 3D assets and scenes.

We work with game studios and creative teams to rethink how professional 3D
workflows are designed in an AI-native world.

The Role

We are looking for a senior generalist software engineer who can own complex
problems across frontend, backend, infrastructure, AI systems, and applied
experimentation.

We do not believe the future of engineering is organized around rigid frontend
and backend boundaries.

As AI reduces the time required to produce code, the engineering bottleneck
shifts toward:

* defining the right problem;
* designing clear system boundaries;
* establishing invariants and constraints;
* validating behavior;
* reviewing generated implementations critically;
* building reusable systems instead of solving every task manually.

You will not only implement features. You will help improve how our team
specifies, generates, reviews, tests, and maintains software.

What You’ll Do

* Own engineering problems from initial definition through production deployment.
* Work across frontend, backend, cloud infrastructure, data, and AI integrations.
* Translate ambiguous product goals into precise behavior, technical invariants, acceptance criteria, and failure cases.
* Use coding agents and LLMs extensively to explore, implement, debug, and review solutions.
* Critically evaluate both AI-generated and human-written code.
* Define what a change may and may not affect, especially in sensitive systems such as billing, permissions, usage limits, and distributed workflows.
* Build automation, internal tooling, evaluation systems, and reusable engineering standards.
* Run applied AI and computer-science experiments and turn successful results into reliable product capabilities.
* Identify recurring development bottlenecks and solve the underlying meta-problem rather than repeatedly fixing individual cases.
* Collaborate directly with product, design, AI research, and customer-facing teams.

What We’re Looking For

* Strong computer-science and software-engineering fundamentals.
* Experience shipping and operating reliable production software.
* Ability to understand systems beyond one narrow technical layer.
* Strong problem decomposition, architecture, and debugging skills.
* Practical experience using AI coding agents or LLMs in serious engineering workflows.
* Ability to define interfaces, invariants, edge cases, failure states, and validation strategies before implementation.
* Comfort moving between rapid experimentation and production engineering.
* Strong judgment about when to automate, simplify, constrain, reuse, or redesign.
* Clear communication and the ability to challenge assumptions constructively.

Particularly Valuable

* Experience with agentic systems, model APIs, evaluation pipelines, or applied machine learning.
* Experience with cloud infrastructure, distributed systems, billing, observability, or developer tooling.
* Familiarity with 3D, games, graphics, creative software, or node-based workflows.
* Experience building internal tools or automation that made an engineering team significantly more effective.
* Strong personal projects, open-source contributions, research, competitions, or technical writing.

How We Work

We use AI aggressively, but not uncritically.

A strong engineer at Atlas does not simply ask an agent to generate a large
pull request. They first define:

* the intended behavior;
* the relevant system boundaries;
* what the change may and may not touch;
* likely failure modes;
* acceptance and validation criteria.

AI is then used to accelerate implementation within those boundaries.

We value people who reason from first principles, take ownership across
conventional role boundaries, improve the development system rather than only
completing tickets, and are willing to discard an implementation when the
underlying specification is wrong.

Example Problems

* Design an agent that creates workflows from a user’s intent while respecting model, cost, and product constraints.
* Add a billing mechanism without unintentionally affecting usage limits, settlements, or existing credit accounting.
* Build automated evaluations for generated 3D outputs.
* Improve a node-based creative canvas across frontend, backend, and model execution.
* Build reusable agent workflows that reduce repetitive implementation and review work.

What Success Looks Like

Within your first months, you will have:

* shipped meaningful capabilities across several parts of the system;
* established clearer technical invariants for critical areas;
* introduced automation that makes the engineering team more effective;
* become trusted not only to implement solutions, but to help determine whether the problem and proposed solution have been defined correctly.

Application Process

Our process focuses on real engineering judgment rather than generic algorithm
exercises.

You should expect:

1. An introductory conversation.
2. A technical discussion with our AI lead.
3. A practical, intentionally ambiguous engineering problem.
4. A discussion of how you would specify it, constrain it, use AI to implement it, and validate the result.

Please include links to relevant GitHub work, technical projects, open-source
contributions, or systems you have shipped.