Spaces:
Running
Running
| title: AMD Enterprise Architecture Optimizer | |
| emoji: ⚡ | |
| colorFrom: red | |
| colorTo: blue | |
| sdk: docker | |
| pinned: true | |
| license: apache-2.0 | |
| app_port: 7860 | |
| # AMD Enterprise Architecture Optimizer | |
| **Powered by AMD MI300X + ROCm | Graph-RAG + DRL + Qwen-72B** | |
| Transform business goals into compliance-grounded EA implementation roadmaps in seconds. | |
| ## What it does | |
| 1. **Graph-RAG Retrieval** — queries 1,400+ enterprise capabilities from Neo4j knowledge graph with vector similarity search | |
| 2. **DRL Prioritization** — REINFORCE policy network (trained on AMD ROCm) prioritises capabilities by business value, budget, and risk | |
| 3. **Qwen-72B Generation** — derives Jira-ready epics, user stories, and acceptance criteria directly from graph-structured governance standards and transformation trends | |
| 4. **Compliance Verification** — self-correcting loop ensures every compliance requirement from TOGAF, ISO 27001, NIST, etc. appears as an acceptance criterion | |
| ## Architecture | |
| ``` | |
| User Input → Graph-RAG Retriever → DRL Optimizer (AMD ROCm) | |
| → Qwen-72B Generator (AMD MI300X via vLLM) | |
| → Compliance Verifier → Roadmap Output | |
| ``` | |
| ## Try It | |
| Select a demo scenario (Healthcare, Banking, Energy) or enter your own organisation profile. | |
| ## AMD Hackathon 2026 — Track 1: AI Agents & Agentic Workflows | |