WITHIN
AI Engineer
About this role
Join our team to architect and develop an internal AI agent platform that powers productivity across the organization. You'll design LLM-integrated applications with tool orchestration, context management, and evaluation frameworks while collaborating with cross-functional teams to expand the platform into document and workflow automation.
What you'll do
- Design and build AI applications leveraging LLM APIs, tool calling, and structured outputs
- Develop agent architectures integrating SQL, Python execution, and marketing platform APIs
- Implement reliability safeguards, retry logic, and hallucination mitigation strategies
- Build evaluation frameworks for LLM systems including multi-turn and regression testing
- Expand platform capabilities into document generation and Google Workspace integrations
- Collaborate with ML engineers, product managers, and full-stack engineers on system reliability
What they're looking for
- Python (production systems)
- LLM APIs (OpenAI, Anthropic, Vertex AI)
- Prompt engineering and multi-turn conversation design
- Tool/function calling and structured outputs
- SQL and database integration
- Cloud platforms (GCP, AWS, Azure)
- API integration and system debugging
- RAG architectures and vector search (plus)
Benefits
- Unlimited vacation policy
- Comprehensive medical, dental, and vision insurance
- 401(k) plan with matching
- Hybrid work opportunity
- Professional development program
- Dog-friendly office and phone stipend
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
WITHIN
WITHIN builds an internal AI agent platform that powers productivity through LLM-integrated applications, tool orchestration, and workflow automation. The company is hiring engineers to architect and develop this platform, expanding its capabilities into document processing and cross-functional automation use cases.
View all jobs at WITHINLikely interview questions
- Walk us through a production LLM application you've built. How did you handle context window limits and multi-turn conversation state?
- Describe your experience designing and implementing safeguards for tool execution—particularly around SQL queries or code execution. What failures have you prevented?