Node.js/Python LLM API Integration

I’m expanding our backend with a brand-new real-time data pipeline that taps directly into large-language-model APIs. The codebase is a mix of Node.js and Python, so the new work must feel native in both environments and follow our existing style.

What I need built
• A service layer that connects to chosen LLM providers (OpenAI, Anthropic, or similar) through their REST or streaming interfaces.
• Real-time processing logic that ingests an internal event stream, sends relevant payloads to the LLM, then pushes the responses back to our clients with sub-second latency.
• Clean, documented endpoints (REST or GraphQL—your recommendation) plus succinct unit tests demonstrating expected behaviour and graceful failure handling.

Acceptance criteria
• End-to-end latency for a single request stays under 1 s for typical 2 k-token prompts.
• Code is lint-clean (eslint / flake8) and ships with at least 90 % test coverage.
• README shows local setup, environment variables, and a cURL example that returns a live LLM response.

I’ll provide access to our Git repo, sample payloads, and LLM API keys as soon as we start.