🧩 Task Distribution & Coordination

• Splitting large models and datasets across multiple GPUs efficiently was complex.
• Ensuring synchronization and consistency in training results was a major challenge.

✅ Trust & Verification

• Verifying that participants actually performed the assigned computation (and did it correctly) required designing a lightweight yet reliable proof system.

🔒 Security & Data Privacy

• Training AI on sensitive data required us to consider privacy-preserving methods like federated learning and secure aggregation.

🌐 Network Latency & Fault Tolerance

• Since we're working with distributed nodes over the internet, managing delays and handling node failures without interrupting training was a tricky part.

🎯 Incentive Mechanism

• Designing a fair and motivating reward system that balances contribution with quality was a challenge, especially without a full blockchain integration.

⏳ Limited Time & Resources

• Being a hackathon project, implementing the complete system in a short timeframe with limited infrastructure was tough, so we prioritized building a functional MVP.