Zof AI is seeking an AI Software Engineer to build our AI products end to end, from the product surface down to the model-facing layer. This role works across the stack: shipping product features, and building the retrieval pipelines, agent orchestration, and context engineering that make AI systems reliable in production. If you have worked as a Software Engineer, AI Engineer, GenAI Engineer, or LLM Engineer, this is that discipline at Zof AI. The ideal candidate is AI-native, has shipped LLM-powered features to real users, and treats quality, cost, and latency as engineering constraints, not afterthoughts.
Engineering · Senior · Full-time · On-site · San Francisco, CA
Responsibilities
- Design, build, and ship production-quality AI product features across the stack.
- Build product features on top of frontier model APIs, including retrieval and RAG pipelines.
- Design agent planning, tool use, and multi-agent orchestration.
- Own prompt and context engineering as a disciplined, tested practice.
- Wire evals into the development loop so quality is measured, not assumed.
- Optimize the cost, quality, and latency of AI features in production.
- Use AI development tools to increase velocity while maintaining quality.
- Own important systems from concept through production.
Requirements
- Strong software engineering foundation.
- Experience building and shipping production systems.
- Experience shipping LLM-powered features to production.
- Working knowledge of retrieval, RAG, and agent patterns.
- Ability to work across the stack.
- Judgment about quality, cost, and latency trade-offs.
- Clear written and verbal communication.
- Comfort operating in a fast-moving environment.
Nice to have
- Experience with TypeScript, React, Next.js, Node.js, Python, Prisma, Postgres, or similar technologies.
- Experience with multi-agent frameworks or agent orchestration systems.
- Experience building or using eval harnesses.
- Experience with test automation, developer tools, or quality systems.
Hands-on experience shipping production features on top of LLM APIs is required
#J-18808-Ljbffr