Software Engineer

Magic’s mission is to build safe AGI that accelerates humanity’s progress on the world’s most important problems. We believe the most promising path to safe AGI lies in automating research and code generation to improve models and solve alignment more reliably than humans can alone. Our approach combines frontier-scale pre-training, domain-specific RL, ultra-long context, and inference-time compute to achieve this goal.

As a Software Engineer at Magic, you will work on core systems or product surfaces that directly determine model capability and user experience.

This role can map onto Pre-training Data, RL Research & Environments, or Product, depending on background and strengths. Across all placements, the expectation is end-to-end ownership: defining problems, implementing solutions, shipping to production, and iterating based on real outcomes.

Magic’s long-context models introduce unique technical challenges — internet-scale data acquisition, long-horizon post-training loops, and product workflows that make complex model behavior understandable and controllable. You will operate close to these constraints, building systems that are both technically rigorous and production-ready.

This role can evolve into deeper specialization in data systems, post-training capability development, or product engineering leadership, depending on strengths and interests.

Depending on team placement, you may:

• Build and scale large distributed data pipelines for pre-training

• Design filtering, mixture, and dataset versioning systems

• Develop post-training datasets, evaluation frameworks, and reward pipelines

• Run ablations that translate capability goals into measurable improvements

• Build end-to-end product surfaces that integrate deeply with the model

• Design APIs, backend services, and frontend workflows for AI-first experiences

• Improve reliability, observability, and performance of production systems

• Strong software engineering fundamentals

• High ownership and comfort operating in ambiguous problem spaces

• Experience building production systems at scale

• Ability to reason clearly about trade-offs between quality, performance, and cost

• Strong technical judgment and bias toward shipping

• Track record of turning complex technical problems into working systems

• Integrity. Words and actions should be aligned

• Hands-on. At Magic, everyone is building

• Teamwork. We move as one team, not N individuals

• Focus. Safely deploy AGI. Everything else is noise

• Quality. Magic should feel like magic