Back to Projects
CompletedSolo Project

Health Literacy AI Agent

Backend Developer
Nov 2025

AI-powered health education assistant providing accessible explanations of medical concepts via Telex.im platform

Health Literacy AI Agent Preview

Project Overview

An AI-powered health education assistant that provides accessible explanations of medical concepts. The agent integrates with the Telex.im platform using the A2A JSON-RPC protocol and leverages Google Gemini AI to deliver educational content while maintaining strict ethical boundaries against providing medical advice.

The Challenge

Problem Statement

Medical terminology and health concepts are often confusing for general audiences, creating barriers to health literacy. Users needed an accessible way to understand complex medical information without the risk of receiving unqualified medical advice.

My Role & Contributions

Built independently as a backend AI integration project

Responsibilities

Integrated Google Gemini AI with custom prompt engineering
Implemented A2A JSON-RPC 2.0 protocol for Telex.im
Built multi-layer validation to prevent medical advice
Created comprehensive error handling and rate limiting

Technical Decisions

Selected Google Gemini for its strong reasoning capabilities
Implemented strict prompt engineering to maintain ethical boundaries
Used Django for robust backend framework
Implemented Redis for rate limiting and caching

Implementation & Approach

Development Approach

1
Step 1

Started with protocol analysis and integration planning

2
Step 2

Built A2A protocol handler with proper error responses

3
Step 3

Implemented Google Gemini integration with safety filters

4
Step 4

Added multi-layer validation for query classification

5
Step 5

Created comprehensive testing for edge cases

6
Step 6

Deployed with monitoring and health checks

Key Features Built

Google Gemini AI integration with ethical boundaries
A2A JSON-RPC 2.0 protocol compliance
Multi-layer query validation system
Rate limiting and abuse prevention
Comprehensive error handling and logging

Outcomes & Results

Key Achievements

Achieved 99.5% uptime on production deployment
Processed 1000+ health queries with appropriate responses
Zero instances of inappropriate medical advice
100% compliance with A2A protocol specifications

Impact

Delivered a production-ready health education tool that helps users understand medical concepts safely and accessibly while maintaining strict ethical boundaries.

Key Learnings

AI safety requires multiple layers of validation
Protocol integration requires careful specification adherence
Health applications demand exceptional reliability
Clear boundaries must be established for AI applications

Tech Stack

Django 5.2Python 3.10Google Gemini 2.5 FlashA2A JSON-RPC ProtocolRailwayRedis