Responsibilities
- Evaluate AI-generated coding interactions end-to-end.
- Judge whether agent outputs are useful, broadly correct, and aligned with strong engineering judgment.
- Assess the quality of explanations, reasoning, preambles, and developer guidance.
- Distinguish different levels of response quality and provide clear, opinionated feedback.
- Identify what worked, what failed, and what felt misleading or off.
- Help define what high-quality interactions look like for tools such as Cursor.
Requirements
- Staff- or Principal-level software engineering experience, or equivalent experience.
- Strong background in TypeScript/JavaScript or Python.
- Hands-on experience using OpenAI Codex, Claude Code, and modern AI-assisted development workflows.
- Ability to evaluate code without fully executing it or reviewing every line deeply.
- Comfort making subjective but rigorous judgments and giving direct, opinionated feedback.
- High standards for engineering quality and judgment.
Benefits
- Contract engagement at approximately 10–20 hours per week.
- Scheduled from ASAP through early May, with a possible extension.
- Selection process includes a take-home evaluation exercise and one behavioral interview.
- Nice-to-have experience includes AI-first IDEs such as Cursor, prompt design or evaluation workflows, mentoring senior engineers, or defining engineering standards.
Tech Stack
Categories
About G2i Inc.
G2I is an AI engineering company that helps the world's leading frontier labs, enterprises, and high-growth startups hire, train, and ship AI systems. Fast. We do three things really well: placing senior engineering talent on demand, building the human workflows and evaluation systems that train and improve AI models, and leading end-to-end product builds for complex engineering challenges. We stay accountable beyond placement, from onboarding and activation through to delivery and retention. That's why our clients tend to see us less as a vendor, and more as a partner in building. 𝗴𝟮𝗶.𝗰𝗼: AI talent on demand. Senior engineers matched and ready in days. 𝗴𝟮𝗶.𝗮𝗶: Engineering expertise powering AI training. We build the environments, rubrics, and human workflows that turn reasoning into structured training data. 𝗼𝗿𝗰.𝗮𝗶: Multi-agent orchestration for complex engineering. Ship code that's scalable, maintainable, tested, and secure. 10 years in. 8,000+ vetted engineers. Clients include Meta, Google, Microsoft, Webflow, Coinbase, 1Password, and the world's leading frontier labs. We also built and run React Miami and AI Engineer Miami, two of the most respected developer conferences in the US. G2I is headquartered in Miami, FL. Learn more at g2i.co, g2i.ai, and orc.ai.