AI Engineer/Senior AI Engineer (Agentic Systems & LLM Integration)

Role: AI Engineer/Senior AI Engineer (Agentic Systems & LLM Integration) Location: Santiago, Chile Employment type: Full Time (Hybrid) Role Summary KR Elixir Technology are hiring an experienced engineer to design, build, and operate production-grade agentic AI systems and LLM integrations. The role owns the technical implementation of multi-agent architectures, retrieval-augmented generation (RAG) pipelines, tool/function calling, evaluation frameworks, and session/state management so that LLM-based applications behave reliably and safely in customer environments. Key Responsibilities • Design and implement multi-agent architectures and runtime components that persist conversation/session state (checkpointers/persistence layers). • Build and operate RAG pipelines, including vector search, reranking (cross-encoder or equivalent), and context hygiene to reduce hallucinations. • Implement function/tool-calling integrations (API/function binding, JSON schema outputs, runtimes that execute and feed results back to LLMs). • Create evaluation and testing frameworks (e.g., LLM-as-a-Judge workflows, golden datasets, rubrics) for automated grading of model outputs. • Design and manage conversation history / context window strategies and session retrieval so multi-turn references resolve correctly. • Work with engineers and infra teams to integrate LLMs, vector DBs, databases (Postgres), and monitoring/telemetry for production resilience. • Produce clear tech designs, runbooks, and unit/integration tests for agent components; enable reproducible failures and component-level observability. • Collaborate closely with product, data science, and QA on deployment, A/B testing, and performance tuning. Required Skills & Experience • Proven experience building production LLM/agent applications or RAG systems. • Hands-on with agent frameworks and persistence concepts (e.g., Checkpointer/MemorySaver/PostgresSaver patterns or equivalent). • Practical experience with vector search and reranking techniques (cross-encoders, two-stage retrieval). • Experience implementing function/tool calling patterns and runtime integrations. • Familiarity with evaluation approaches for LLMs (LLM-as-a-Judge or programmatic evaluation pipelines). • Strong knowledge of session state, context windows, and chat history management. • Comfortable with backend systems: REST/GraphQL APIs, Postgres (or equivalent), vector DBs, and cloud deployments. • Solid software engineering fundamentals: testing, CI/CD, observability, and fault isolation. • Excellent debugging and systems-thinking skills to locate failures at component level. Preferred / Nice-to-have • Experience with specific agent/LLM frameworks (LangGraph, LangChain, AutoGen, etc.). • Familiarity with cross-encoder models, embedding pipelines and popular vector stores. • Prior work on production monitoring for LLM-based apps (latency, cost controls, hallucination metrics). • Background in designing human-in-the-loop fallbacks and safety/guardrails for autonomous agents.