About this role
The AI Engineer builds and operates the agentic workflow that powers ReaderSight — the multi-agent design that harvests signals, triggers alerts, and lets buyers query titles conversationally. This role owns the Claude LLM integration and the automated weekly scoring run end to end.
- Build the multi-agent workflow: Signal Harvesting Agent, Scoring Agent, Alert Agent, and Conversation Agent
- Implement Agent Bricks orchestration within the existing Databricks environment
- Integrate and tune the Claude LLM harness for sentiment, entity, and narrative extraction from messy text and video transcripts
- Build the conversational query layer that lets buyers ask plain-English questions about any title or category
- Implement threshold-triggered alert generation with per-signal evidence summaries for flagged titles
- Build the automated weekly scoring run across all titles with an on-sale date within the 90-day window
- Support the backend logic for the Summary UI (ranked list, evidence drill-down, alert dashboard)
- Participate in LLM harness and scope testing during UAT