Intelligent Distributed Compute + Storage System
Automatically share computing power and storage across machines — with zero manual effort.
Zero Compute is a lightweight distributed system that transforms multiple machines into a single intelligent computing environment.
It not only shares compute power across devices but also manages storage dynamically — automatically moving, syncing, and retrieving data based on usage patterns.
Your system behaves like one machine — regardless of where the data or compute lives.
Access File / Run Task
↓
System decides:
- Run locally?
- Offload compute?
- Fetch file from another machine?
↓
Execute → Return result transparently
Unlike traditional systems:
- No manual file transfer
- No manual compute distribution
- No explicit remote execution calls
Zero Compute introduces intelligence into both compute and storage layers.
- Automatically offloads heavy tasks to other machines
- Uses available CPU resources across the network
- Falls back to local execution if needed
- Automatically shifts rarely used files to other machines
- Keeps frequently accessed files local
- Pulls files back instantly when needed
- Users interact with files and programs normally
- System handles location, transfer, and execution internally
flowchart LR
A[Primary Machine] -->|Query Nodes| B[Control Service]
B -->|Node Info| A
A -->|Offload Task| C[Worker Node]
C -->|Execute| D[CPU Execution]
D -->|Result| C
C -->|Response| A
A -->|Request File| E[Remote Storage Node]
E -->|Send File| A
| Component | Description |
|---|---|
| Primary Agent | Controls task execution and file access decisions |
| Worker Node | Executes compute tasks from other machines |
| Storage Node | Stores files and serves them on demand |
| Control Service | Tracks nodes and manages discovery |
-
User opens a file or runs a task
-
System checks:
- Where is the file?
- Is the task heavy?
-
If file is remote → it is fetched automatically
-
If task is heavy → it is offloaded
-
Execution happens
-
Result is returned seamlessly
- Frequently used files → stay local
- Rarely used files → moved to other machines
- Large tasks → distributed automatically
- System adapts based on usage patterns
zero-compute/
│
├── control-service/ # Node registry
├── worker-node/ # Compute execution
├── storage-node/ # File storage + transfer
├── primary-agent/ # Decision engine
├── shared/ # Configs & schemas
└── docs/ # Documentation
- CPU-based compute sharing
- Basic file transfer between machines
- Simple usage-based file movement
- Single worker and storage node
- GPU-based compute sharing
- Smart caching algorithms
- Predictive file placement
- Multi-node scheduling
- Distributed file system layer
- Cross-network support
- Extend laptop performance using other machines
- Share storage across personal devices
- Run heavy tasks without upgrading hardware
- Build a personal distributed system
Compute and storage should not be tied to a single machine.
Zero Compute creates a system where resources are shared, adaptive, and invisible to the user.
Contributions are welcome. Focus areas include scheduling, storage strategies, and distributed coordination.
MIT License