A production-style multi-agent AI system designed to generate, validate, and automate structured social media engagement workflows with human-in-the-loop control.
This project demonstrates my ability to design scalable AI automation architectures using multi-agent orchestration, structured outputs, and reliable workflow management.
Social media engagement is repetitive and time-consuming.
This system:
- Automates AI-based comment generation
- Keeps human approval before publishing
- Ensures safe and ordered posting (FIFO queue)
- Tracks engagement metrics for continuous improvement
- Implements robust logging and fallback mechanisms
The platform is divided into three layers:
- Built with React / Next.js / TypeScript
- User authentication
- Social account connection
- Comment review & approval interface
- Node.js / Express
- MongoDB
- Handles workflow logic
- Manages comment lifecycle
- Controls multi-agent coordination
- Python microservices
- LLM integration (Ollama or other models)
- Comment generation
- Engagement analysis
- Context-aware outputs
- FIFO-based publishing queue
- Ensures ordered and safe comment deployment
Each connected social platform uses:
- Generates contextual comments
- Produces structured outputs
- Sends suggestions for human validation
- Manages ordered publishing
- Ensures safe workflow execution
- Tracks engagement metrics
- Monitors likes, replies, and performance
- Feeds data back into the system
- User registers / logs in
- Connects social account
- Selects target accounts or hashtags
- AI generates contextual comments
- User approves / edits / rejects
- Approved comments are published via FIFO queue
- Engagement metrics are analyzed
- React / Next.js
- Node.js
- Express
- TypeScript
- MongoDB
- Python
- LLM (Ollama)
- Multi-agent orchestration
- FIFO queue system
ts-node src/server.ts
(Include full setup instructions here if needed.)
🔮 Future Improvements
Intelligent scheduling for queue optimization
Advanced analytics dashboard
Cross-platform engagement insights
Continuous learning from engagement feedback
Developed by Kawtar CHAKIR
AI/ML Engineer specializing in multi-agent systems and AI automation.