Role Overview
We are seeking experienced Software Engineers to create challenging Reinforcement Learning (RL) environments that evaluate the ability of advanced AI models to solve real-world software engineering problems. In this role, you will design coding challenges based on real software repositories, build reproducible development environments, and create reliable verification systems and reference solutions.
Responsibilities
- Create realistic and challenging coding tasks for advanced AI models using repositories you know well.
- Design problems that require genuine engineering expertise, including subtle bug fixes, non-trivial feature implementation, and meaningful performance improvements.
- Build reproducible environments for each coding task.
- Develop deterministic verifiers that accept correct solutions across different valid implementations while rejecting incorrect, incomplete, or hard-coded approaches.
- Write reference solutions demonstrating how each task can be successfully completed.
- Ensure tasks are technically sound, clearly defined, and independently verifiable.
- Identify and flag tasks that are ambiguous or cannot be reliably verified rather than submitting them.
- Document technical requirements and communicate problem specifications clearly.
Requirements
- Strong understanding of algorithms, data structures, and software engineering principles.
- Proven ability to debug complex systems and resolve challenging software issues.
- Experience implementing features, refactoring code, and optimizing application performance.
- Ability to build reproducible development environments and write reliable verification tests.
Preferred Qualifications
- Experience working with large codebases and modernizing legacy systems.
- Track record of delivering high-impact software features from conception through implementation.
- Strong technical documentation and communication skills.
- Interest in artificial intelligence and its technical challenges.