🎯 Project #1: Local Multi-Provider Chat Agent (Ollama-first)

Source code: /chat-agent-api

Build a simple HTTP service that:

Uses Ollama as the primary model (e.g., llama3), but keeps the code ready to switch providers via init_chat_model-style config.

Exposes:

• Single-turn Q&A (/ask).
• Multi-turn chat with message history using HumanMessage and AIMessage.
• Wraps everything in a small "agent" abstraction (no tools yet) using create_agent so you can later add tools.

Example usage flow:

POST /chat with { 
  "session_id": "abc", 
  "message": "Tell me about Luna City" 
}

• Service loads past messages for session_id, calls the agent, streams tokens back to the client.

• All LLM calls go through Ollama instead of OpenAI.

🎯 Project #2: YouTube Content Researcher API

Source code: /youtube-researcher-api

100% Local Ollama - no external APIs

What it does:

• /research → Analyzes video topics, suggests titles/thumbnails
• /trending → Finds hot YouTube niches (AI/IoT/embedded)
• /competitors → Analyzes competitor channels