Summary

The SMB AdTech Platform offers AI-powered, privacy-compliant programmatic advertising infrastructure tailored for U.S. small and medium-sized businesses. It democratizes access to enterprise-grade advertising tools, enabling SMBs to overcome limitations of "walled garden" platforms, technical barriers, and privacy regulations. The platform provides a self-serve, no-code interface with AI assistance, allowing non-technical users to create and optimize campaigns across a broader range of inventory, maximizing ROI through automated, real-time bidding.

How It Works

The platform utilizes a sophisticated, multi-model Machine Learning stack for real-time bidding and optimization. It employs DeepFM for predicting click-through rates, Graph Attention Networks (GAT) to model relationships and discover audience segments, and a Proximal Policy Optimization (PPO) reinforcement learning agent for dynamic bid adjustments and budget pacing. This approach allows for bid decisions in under 50 milliseconds. Privacy-first attribution is handled via probabilistic models and synthetic data, ensuring compliance with regulations like Apple's App Tracking Transparency (ATT) without relying on user-level identifiers.

Quick Start & Requirements

• Primary install/run command: Clone the repository, install core dependencies (pip install -r requirements.txt), and run the API (uvicorn api.main:app --reload). Docker Compose is also provided (docker-compose up -d).
• Prerequisites: Python 3.11+, Docker & Docker Compose. Redis is optional. PyTorch is recommended for full ML capabilities (DeepFM, GAT, PPO); without it, the system gracefully degrades to scikit-learn GBM.
• Links: GitHub repository: https://github.com/haoyangfeng2024/smb-adtech-platform.git. API documentation is available at http://localhost:8000/docs after running the service.

Highlighted Details

• Cross-Publisher Delivery: Access to inventory across the open internet, mobile apps, and CTV, extending reach beyond major platforms like Meta and Google.
• Privacy-First Attribution: Probabilistic models and synthetic data generation ensure compliance with ATT and U.S. state privacy laws without user-level tracking.
• LLM Marketing Assistant: Enables non-technical users to generate ad copy and optimize campaigns through natural language prompts.
• Automated ML Bidding: Real-time bid optimization leveraging DeepFM, GAT, and PPO, with fallback mechanisms to GBM and heuristic rules.
• Explainable AI (XAI): Features like a live "Bidding Curve" showing PPO agent bid adjustments (δ) and ROI trend analysis offer transparency into AI optimization strategies.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), sponsorships, or partnerships are provided in the README.

Licensing & Compatibility

• License Type: MIT License.
• Compatibility Notes: The permissive MIT license generally supports commercial use and integration with closed-source applications.

Limitations & Caveats

The frontend dashboard and LLM assistant integration are explicitly marked as "Work In Progress" (WIP). The roadmap indicates further development is planned for multi-SSP supply integration and real-time reporting dashboards.