Updated description (aligned to “ChatGPT-style app + backend + API + browser UI + cURL”)

MC02: What is Generative AI? | GitHub Basics | AI App Architecture (ChatGPT-Style)
Part of AI Residency Cohort 9 (Fast Track)

In this module, we’ll break down a modern ChatGPT-style AI application in the simplest, most practical way: Frontend UI → Backend API → LLM Provider → Response back to the browser, with the same backend also testable via cURL and other API clients.

The AI App Architecture (the real-world view)

You’ll understand how an AI app works across four entry points:

1. Browser UI (Chat Interface)

• A chat front end (web) where the user types messages
• The UI sends a request (usually JSON) to your backend endpoint

1. Backend (API Server)

• Your server (e.g., FastAPI / Node) receives the request
• It applies rules: prompt template, system instructions, safety checks, context formatting
• It calls the LLM provider (OpenAI / Bedrock / open-source gateway)
• It returns the final response to the UI

1. API-first usage (cURL / Postman)

• Same backend endpoint can be called from terminal:
• Useful for debugging, testing latency, and sharing reproducible requests
• You learn how to validate your app without relying on the UI

1. Direct API consumption (Integrations)

• Other apps/services can call your endpoint programmatically
• This is how AI features get embedded into products, workflows, and automation

What we will cover in MC02

• What GenAI is (and how it differs from classic ML and rule-based automation)
• The end-to-end request flow: Browser UI → Backend → LLM → Response
• Core components you must have
• Prompt + system instructions
• Input/output schema (clean JSON contracts)
• Error handling + timeouts
• Basic guardrails (what the app can/can’t do)
• GitHub basics for AI apps
• Repo structure for frontend/backend
• README that explains setup + API usage
• Commit discipline so your project is portfolio-ready

Outcome

By the end, you will be able to clearly explain (and start building) a ChatGPT-like AI app that works:

• in the browser (UI)
• via API calls (cURL/Postman)
• with a clean backend architecture you can extend into RAG, tools, and agents later

Join AI Residency Cohort 9 (Fast Track):
https://academy.decodingdatascience.com/airesidencyfasttrack