We are looking for a senior AI / backend engineer to work on a production-grade vision + LLM inference pipeline. This role is focused on system design and implementation, not experimentation or prompt-only work. The system combines: Vision-enabled LLMs Optional secondary models (OpenAI fallback) Backend policy and validation layers Progressive / staged responses for low-latency UX This is a technical role for someone who understands how AI models behave in production, including failure modes, cost tradeoffs, and latency constraints. Scope of Work You will help design and implement: Vision-based image analysis Multi-call LLM pipelines (vision + text generation) Parallel and progressive inference flows Backend validation and safety filters Prompt control via system vs runtime instructions Model routing and fallback logic Token, latency, and error observability Tool / function calling integration where appropriate Required Skills You should be strong in most of the following: Python (FastAPI or similar backend frameworks) Vision-enabled LLMs (Groq, OpenAI, or comparable) Designing LLM pipelines, not just single prompts Parallel execution and streaming responses (SSE, NDJSON, or equivalent) Prompt control and structured outputs Guardrails for hallucination, safety, and coherence Performance optimization for inference latency and cost Clean, maintainable backend architecture Strong Plus If You Have Experience with LLM tool / function calling Prior work on image understanding systems Familiarity with AWS or similar cloud environments Experience debugging real-world LLM failures What This Role Is Not Not prompt engineering only Not chatbot development Not academic research Not a one-off script This is systems-level AI engineering.