AI Integration Engineer

About the role

The AI Integration Engineer is responsible for connecting large‑language‑model (LLM) APIs with existing product systems and delivering AI‑driven features to end users. This role focuses on reliable, secure, and scalable integration rather than model training.

Key responsibilities

• Integrate LLM APIs such as Anthropic Claude, OpenAI GPT‑4o, AWS Bedrock, and Google Gemini into backend services and user‑facing products.
• Design and implement Retrieval‑Augmented Generation (RAG) pipelines, including document ingestion, chunking, vector store selection, and retrieval tuning.
• Build agentic workflows using frameworks like AgentCore, LangChain, LlamaIndex, or custom orchestration patterns.
• Manage prompt engineering, versioning, and evaluation frameworks.
• Implement guardrails for LLM outputs, including validation, content filtering, and fallback logic.
• Monitor AI system performance (latency, cost‑per‑query, accuracy drift, token usage) and ensure observability in production.
• Collaborate with frontend engineers to expose AI capabilities in product UIs.
• Own the AI integration layer throughout the software development lifecycle, from specification to CI/CD and production monitoring.

Required profile

• 3+ years of backend or full‑stack development experience with strong API design skills.
• Proven experience integrating LLM APIs into production applications.
• Proficiency in Python and/or TypeScript/Node.js.
• Hands‑on experience with RAG pipelines, vector databases, and embedding models.
• Solid understanding of prompt engineering techniques.
• Knowledge of API security, rate limiting, and cost management for LLM services.
• Experience with AWS or another major cloud platform.

Required skills

• Python
• TypeScript / Node.js
• LLM API integration (Anthropic, OpenAI, AWS Bedrock, Google Gemini)
• RAG pipeline development
• Vector databases (Pinecone, Weaviate, pgvector)
• Embedding models and chunking strategies
• Prompt engineering
• API security, rate limiting, cost management
• AWS (or equivalent cloud services)
• Agentic frameworks (AgentCore, LangChain, LlamaIndex, CrewAI, AutoGen)
• Multi‑modal AI (vision, audio) and function calling
• Fine‑tuning workflow awareness
• AI coding assistants for development acceleration