AI Engineer - Remote

You'll build and optimize our AI travel chat: a contextual, lightning-fast assistant that works seamlessly alongside our visual travel planner and booking product. The chat needs to understand travel context deeply — destinations, logistics, preferences, partner constraints
and translate that into actionable itinerary changes in real-time. Speed and responsiveness are non-negotiable.

- Design and implement a multi-agent system (orchestrator + specialist worker agents) for contextual travel planning, using frameworks like LangGraph or Mastra
- Build a dual-state architecture: a JSON-based AI state (backend source of truth) synced to the interactive UI state (maps, calendars) via a framework like CopilotKit
- Develop and optimize a RAG pipeline to ingest unstructured B2B partner knowledge (PDFs, offers, tone-of-voice guidelines) into actionable constraints for the AI
- Implement a POI/location caching strategy with smart TTL to minimize Google Places API costs while keeping data fresh
- Wrap existing GraphQL API endpoints as agent-callable tools (function calling) so AI can interact with current Stippl functionality without breaking it
- Set up evaluation pipelines (LLM-as-a-judge), input/output guardrails, and observability (e.g. Langfuse) to ensure quality, safety, and continuous improvement
- Optimize for speed: low-latency responses, efficient token usage, and the right LLM per agent task

Our architecture follows a Data Core + AI Core + Output model. The Data Core handles ingestion from multiple sources into a RAG framework with vector storage. The AI Core runs an orchestrator agent that delegates to specialist workers (client profiling, curation, logistics, critic agents) with function calling tools. Output is a generated/adjusted travel plan rendered in both the Stippl Planner and chat interface. You’ll be working across this full stack with a focus on the AI Core and its integration points.

- Proven experience building production AI/LLM applications — not just prototypes. You’ve shipped agentic systems, RAG pipelines, or AI-powered products before
- Strong backend engineering skills (Node.js/TypeScript or Python). Our stack includes GraphQL, MongoDB, and we’re evaluating vector databases like Weaviate
- Deep understanding of LLM orchestration: prompt engineering, multi-agent coordination, context management, and function calling
- Experience with relevant frameworks: LangGraph, LangChain, Mastra, CopilotKit, or similar
Obsession with performance — you know how to make AI interactions feel instant through streaming, caching, model selection, and architecture choices
- Familiarity with evaluation and guardrailing patterns for LLM systems
- Bonus: travel industry experience or domain knowledge

- Receive 100% salary from the onboarding date.
- Participate in company activities: Teambuilding, travel, vacation and other activities.
- Work with large and advanced systems, have the opportunity to develop comprehensive technology skills with complex problems, requiring high accuracy.