AI Rust Engineer at Zed Industries building AI systems that operate inside evolving codebases. Focus on designing context systems, evaluation frameworks, and infrastructure connecting language models to the editor and developer tools.
Software development is becoming a collaboration between developers and AI systems.
Most AI developer tools operate on prompts and static snapshots of code. But real software development happens inside large, evolving codebases that change continuously through edits, refactors, discussions, and collaboration.
We are building AI systems that operate inside that environment.
Zed today is a code editor built from scratch in Rust with GPU acceleration and real-time multiplayer collaboration. Every keystroke and edit can be shared instantly between collaborators, creating a shared workspace where developers and increasingly AI systems can work together.
We are also building DeltaDB, a system that records the operational history of software development. It captures edits, discussions, and AI interactions as they happen. This creates a foundation for AI systems that can reason not just about code, but about how that code evolves over time.
Much of this work involves designing systems that help AI understand large, evolving codebases. This includes providing the right context and evaluating whether its suggestions actually improve the software.
The work sits at the intersection of AI systems, developer tools, and distributed collaboration. We are exploring how software gets built when AI becomes part of the development environment itself.
Many of these questions do not yet have established answers. Part of the work is designing and testing new approaches in real developer workflows.
How should AI reason about codebases that are constantly evolving rather than static repositories?
How can streams of development activity such as edits, refactors, and discussions provide useful context for models?
What does it mean for AI to operate inside a real-time collaborative environment where multiple developers are editing the same project?
How should AI systems make suggestions without interrupting developer flow?
How do we evaluate whether AI systems are genuinely helping developers understand and evolve complex software systems?
Designing systems that allow AI to participate directly in developer workflows
Building infrastructure that connects language models with the editor and developer tools
Developing context systems that help models reason about large codebases
Designing evaluation frameworks for AI-assisted development
Improving the reliability, latency, and cost efficiency of AI features
Working closely with editor and infrastructure engineers to ship ideas quickly
Pair programming with teammates to explore ideas and refine systems together
Experience building production systems powered by large language models
Strong understanding of model behavior, prompting, and evaluation
Experience integrating AI capabilities into real software products
Strong backend or systems programming experience
Interest in developer tools and programming environments
Ability to collaborate closely with other engineers
Experience with Rust, or willingness to learn
Building AI coding assistants or developer tools
Designing evaluation systems for LLM-driven products
Working with large codebases, compilers, or programming environments
Experience with tool-using or agent-style models
Experience with Rust
Zed is open source and built in public by the team behind Atom and Tree-sitter.
The editor is written in Rust with GPU acceleration for every frame. When you type or move the cursor, pixels respond instantly. That responsiveness keeps you in flow.
Zed is multiplayer by default, allowing developers to work together in the same codebase in real time. Much of our work happens through pair programming, with engineers collaborating directly inside the editor.
We ship improvements weekly and work closely with a community that cares deeply about the craft of developer tools.
Location: Remote (American or European time zones)