MedArchitect AI

From medical-device requirements to a grounded system architecture.

MedArchitect AI is an AI-assisted medical-device system design platform built during a hackathon.

I wanted to explore a simple question:

Can an LLM be made more useful for engineering by grounding it in the right knowledge instead of relying only on what it learned during training?

Overview · How It Works · Architecture · Development

---

Overview

A general LLM can generate a convincing medical-device design, but it can also hallucinate technical requirements, standards, or other information.

MedArchitect AI takes a different approach.

It maintains a knowledge base of medical-device technical and regulatory documents, converts them into searchable embeddings, retrieves the most relevant information for a user's requirements, and provides that context to a local LLM before generating the design.

User Requirements
        │
        ▼
   RAG Pipeline
        │
        ├─────────────── Knowledge Base
        │                       │
        │                       ▼
        │                 PDF Documents
        │                       │
        │                       ▼
        │                  Embeddings
        │                       │
        │                       ▼
        │                    ChromaDB
        │                       │
        └───────────┬───────────┘
                    ▼
             Relevant Context
                    │
                    ▼
              Gemma 3:1B
                    │
                    ▼
           Structured Design
                    │
                    ▼
                Web UI