AI RAG Decision Platform
Publicada el 2026-07-21
Descripción de la oferta
I need a Retrieval-Augmented Generation (RAG) platform that can analyse live and historical data streams, then trigger automated, real-time operational decisions without human intervention. The core requirement is straightforward: feed the system structured and unstructured data, receive concise Red-Amber-Green recommendations or direct actions that keep day-to-day operations running smoothly. Key elements I expect you to handle include: • Data pipeline: ingest multiple internal data sources via API or secure file drop, normalise them, and store them for low-latency retrieval. • RAG engine: combine a vector database with an LLM of your choice (e.g. OpenAI GPT-4, Llama-2, Claude) so the model can ground every answer in the most relevant documents. • Decision layer: map the model’s output to deterministic business rules and produce a final action flag (red, amber, green) plus an explanation. • Dashboard & API: surface results through a lightweight web UI and a REST endpoint so other systems can consume them. Success criteria 1. Given a set of fresh operational inputs, the platform returns a colour flag in <3 seconds. 2. The supporting evidence is traceable to source documents. 3. I can retrain or extend the knowledge base through a simple upload or API call. Deliver a runnable solution (Docker-ised or cloud deployable), well-commented code, and brief documentation so I can keep refining the rules on my side.
Skills
Fuente original: freelancer