{"@context":"https://schema.org","@type":"JobPosting","title":"AI Agent Engineer (LLM Orchestration & Autonomous Systems)","description":"\n

Build the autonomous agent infrastructure that powers Wiktis. You'll architect multi-agent systems, design tool-use patterns, implement memory and retrieval layers, and deploy production-grade AI agents that handle complex exhibition workflows—from exhibitor onboarding to floor plan optimisation to automated reporting. This is a specialised role for engineers who understand agent reasoning loops, structured outputs, and the orchestration layer that turns LLMs into reliable autonomous systems.

\n \n

Responsibilities:

Architect and implement multi-agent systems using LangGraph, CrewAI, AutoGen, or custom orchestration frameworks
Design agent tool schemas, function calling interfaces, and structured output parsers
Build agent memory systems: short-term context windows, long-term vector retrieval, and episodic memory patterns
Implement RAG pipelines with hybrid search (semantic + keyword), reranking, and contextual chunking strategies
Develop agent evaluation frameworks: task completion metrics, hallucination detection, and regression testing
Create agent observability systems: trace logging, token usage tracking, and latency monitoring
Design fallback strategies, human-in-the-loop handoffs, and graceful degradation patterns
Implement guardrails, output validation, and safety layers for production agent deployment
Build prompt templates, chain-of-thought reasoning patterns, and few-shot example libraries
Collaborate with product to translate user workflows into agent task decomposition

\n \n

Qualifications:

Deep experience with agent frameworks: LangChain, LangGraph, LlamaIndex, CrewAI, or AutoGen
Strong understanding of LLM capabilities and limitations: context windows, token economics, and model selection
Proficiency with function calling, tool use, and structured output formats (JSON mode, Pydantic models)
Experience building RAG systems: embedding models, vector databases, chunking strategies, and retrieval evaluation
Knowledge of prompt engineering patterns: chain-of-thought, ReAct, tree-of-thought, and self-consistency
Backend development skills in Python or TypeScript for agent infrastructure
Understanding of agent memory architectures and conversation state management
Experience with LLM observability tools: LangSmith, Weights & Biases, Helicone, or custom tracing

\n \n \n

Nice to Have:

Experience with multi-agent coordination and agent communication protocols
Knowledge of fine-tuning, RLHF, or model distillation for task-specific agents
Background in reinforcement learning or planning algorithms
Familiarity with voice agents, multimodal models, or real-time agent interfaces
Experience deploying agents at scale with rate limiting, caching, and cost optimisation
Contributions to open-source agent frameworks or LLM tooling

\n \n \n

Benefits:

Remote-first culture
Top-tier compensation and equity
Comprehensive health benefits
AI tools, API credits, and model access budget
Conference attendance (NeurIPS, AI Engineer Summit, LLM conferences)
Research time allocation for agent experimentation
Flexible schedule
Latest development hardware

\n ","datePosted":"2026-02-04","validThrough":"2026-05-04","employmentType":"FULL_TIME","hiringOrganization":{"@type":"Organization","name":"Wiktis","sameAs":"https://www.wiktis.com","logo":"https://www.wiktis.com/brand/logos/Wiktis-Symbol.webp"},"jobLocation":{"@type":"Place","address":{"@type":"PostalAddress","addressLocality":"Remote","addressCountry":"US"}},"baseSalary":{"@type":"MonetaryAmount","currency":"USD","value":{"@type":"QuantitativeValue","minValue":110000,"maxValue":160000,"unitText":"YEAR"}},"jobLocationType":"TELECOMMUTE","applicantLocationRequirements":{"@type":"Country","name":"US"}}