Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
76 changes: 76 additions & 0 deletions Project_Plan.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,76 @@
# Production-Ready Medical AI Assistant

**Author:** Precious T.

---

## Project Overview
This project builds a medical AI assistant using **Python**, **FastAPI**, **Tavili**, **Gemini**, and **LangChain**.
It’s designed to answer patient questions, suggest possible diagnoses, and pull in evidence-based info from trusted APIs.
The system is modular and ready for use in a real-world production environment.

---

## Architecture Diagram
```
[Client / UI]
[FastAPI Server] ── Authentication & Request Validation
├─> [LangChain Pipeline] – Prompt Templates, Memory, Tools
├─> [Gemini Model] – Medical language generation
├─> [Tavili Vector DB] – Store & retrieve medical knowledge
[Response JSON to Client]
```
---

## ⚙️ Main Components

### FastAPI Endpoints
- `POST /ask` – Accepts user medical questions
- `GET /history` – Retrieves past conversations
- `POST /feedback` – Collects user feedback for ongoing improvements

### LangChain Integration
- Uses prompt templates to structure medical queries
- Keeps conversation context with memory

### Gemini Model
- Core language model generating answers
- Fine-tuned specifically for safe medical information

### Tavili Vector DB
- Stores trusted medical documents
- Allows semantic search to enrich answers with relevant knowledge

### Security & Compliance
- Basic authentication and API key protection
- Designed to handle data carefully, similar to HIPAA standards (no logging of personal health info)

---

## Implementation Plan

- Build the FastAPI structure and endpoints
- Connect LangChain and Gemini for answering questions
- Integrate Tavili for knowledge retrieval and context handling
- Add authentication, logging, and deploy to production

---

## Deployment
- Package the app using **Docker**
- Deploy securely to platforms like Render or Heroku with **HTTPS**
- Manage secrets through environment variables

---

## Current Status
Due to a power outage, I’ve completed the architecture and design so far.
Full code will be pushed as soon as power is back.
I’m also open to teaming up with a teammate tonight to speed things along.