AEGIS

🧭 Inspiration

We live in a world where data is the heartbeat of every mission — whether it’s F1 race telemetry streaming in milliseconds or life-saving medical data being uploaded from a disaster zone. We saw one recurring problem: data transfers fail when it matters the most.

When a 50 GB surgery video stops at 99%, or when drone coordinates vanish mid-flight, the damage isn’t just technical — it’s human. So, we asked a simple question: “Why rely on one unstable link when your data can adapt, think, and survive on its own?”

That’s how AEGIS was born — an intelligent, self-healing file transfer system designed for the unpredictable.

⚙️ What It Does

AEGIS ensures your data always gets delivered — no matter the network chaos. It breaks files into smart chunks, predicts link failures before they occur, and dynamically reroutes each chunk using the best available channel — Wi-Fi, 5G, Bluetooth, or Mesh.

Even if the internet dies, AEGIS builds its own local network — peer-to-peer — to keep the mission alive. It’s a file transfer system that doesn’t just move data; it thinks, adapts, and protects it.

🧩 How We Built It

Frontend: React.js dashboard for real-time transfer monitoring.

Backend: Python for adaptive control logic and AI orchestration.

AI Engine: TensorFlow Lite model predicting network failures up to 30 seconds ahead.

Networking: WebRTC for peer-to-peer adaptive file routing.

Decentralization: IPFS layer for distributed, fault-tolerant storage.

Databases: Redis for live transfer caching and PostgreSQL for persistent metadata.

Each layer talks to the others through lightweight APIs — forming a three-tier defense:

Resilient Core: Chunking + blockchain-style verification.

AI Channel Orchestrator: Dynamic, predictive routing.

Mesh Fallback: Offline peer-to-peer continuity.

🧱 Challenges We Ran Into

Keeping data synchronized while chunks switched between networks mid-transfer.

Preserving integrity without slowing performance.

Ensuring seamless fallback to mesh without duplicate transfers.

Building a UI that could visualize complex transfers simply and clearly.

🏆 Accomplishments We’re Proud Of

Built a system that recovers from network loss automatically — no manual restarts.

Achieved predictive rerouting using lightweight ML under 30 ms inference.

Integrated multi-protocol and P2P layers smoothly within one architecture.

Designed a dashboard that shows live AI decisions in action — visually and intuitively.

📚 What We Learned

We discovered how fragile existing transfer protocols truly are — and how powerful adaptive systems can be when intelligence meets networking. We learned the delicate balance between speed, redundancy, and reliability, and how machine learning can turn prediction into prevention.

Mathematically, AEGIS stands on a simple principle: ( \text{Reliability} = 1 - \prod_{i=1}^{n} (1 - p_i) )

​

is the success probability of each independent path — proving that redundancy exponentially increases reliability.

🚀 What’s Next for AEGIS

Deploying an edge-based version for autonomous vehicles and remote healthcare devices.

Training the ML model on real-world latency datasets to enhance prediction accuracy.

Adding AES-256 end-to-end encryption for enterprise-grade security.

Building a mobile app version for instant disaster-response transfers.

AEGIS isn’t just a hackathon project — it’s a step toward the future of self-healing data transmission.

Built With

• grafana
• ipfs
• java
• lite
• postgresql
• python
• react.js
• redis
• tensorflow
• webrtc