Back to work
Case 06 · AI Engine · Mobile · 2026

Multi-LLM powered
clinical reasoning.

A mobile health assistant app orchestrating OpenAI GPT, Anthropic Claude, and Google Gemini APIs. Integrating modern front-end design with strict backend data flows to deliver a reliable, secure user experience.

Loading scene
Role
Full-Stack Mobile Developer
Year
2026 - Ongoing
Domain
Healthcare AI · Mobile Applications
Platform
iOS · Android
Core stack
Unity (C#) · TypeScript · REST APIs
Backend
Firebase Auth · Firestore
Overview

Routing queries to the best mind.

Different Large Language Models excel at different tasks. Some are better at parsing raw medical texts, while others are superior at conversational empathy or image analysis. The challenge was building an architecture that utilizes the right model for the right task dynamically.

MEDIBOT is an AI-powered Turkish health assistant that sits on top of a sophisticated TypeScript backend routing layer. It seamlessly connects a Unity C# mobile client to OpenAI, Anthropic, and Google Gemini via strictly managed REST endpoints.

With an integrated 3-tier subscription system, robust Firebase authentication, and a real-time database, it ensures patient-level data remains secure while delivering instant, multi-model AI responses.

Process

Architecting the brain.

01
Backend orchestration

Structuring the TypeScript middleware layer to securely handle API keys and format diverse LLM payloads.

02
REST integration

Building an async/await data pipeline in Unity (C#) to handle JSON parsing without dropping framerates.

03
Firebase ecosystem

Hooking up Firestore for cross-session chat persistence and Firebase Auth for secure user onboarding.

04
Monetization setup

Deploying a dynamic 3-tier subscription model logic that controls prompt token limits and model routing.

Capabilities

Under the hood.

Multi-LLM Routing

Intelligently dispatches user queries to OpenAI, Anthropic, or Google APIs based on the query context and the user's subscription tier.

Async REST architecture

Non-blocking network requests from Unity using modern C# async/await patterns, ensuring the mobile UI remains smooth during high-latency LLM generations.

Image analysis pipeline

A unified pipeline allowing users to upload clinical documents or images, converting them into secure payloads ready for multi-modal AI vision processing.

Firebase infrastructure

A resilient real-time database architecture enabling instant state synchronization across devices, backed by enterprise-grade user authentication.