Paai is building an AI-powered platform where agents execute complete professional workflows end-to-end. The role involves being the technical leader for the AI platform, focusing on agent orchestration, search and retrieval infrastructure, and model integration.
Responsibilities:
- Build the core agent loop: intent classification, tool-use and function-calling pipelines, multi-step workflow execution, and result synthesis
- Design the layer that connects agents to knowledge across internal databases, external search APIs, research tools, and structured data providers
- Platform-level systems for session state, entity tracking, and memory across multi-step tool chains and follow-up conversations
- Infrastructure for onboarding models and routing to the right one per task
- Rendering infrastructure where multi-step agent work streams progressively, structured documents build in real-time, and deliverable cards appear as each phase completes
- The routing system that maps hundreds of natural language phrasing variations to the correct workflow
- Golden datasets, automated scoring, regression detection