TIM: The Inference Mesh

"Some call me... Tim."

---

A peer-to-peer network for sharing spare inference capacity. Think BitTorrent for intelligence. SETI@home for inference.

Network Topology

┌────────────────────────────────────────────────────────────────────────────┐
│                              TIM MESH NETWORK                              │
├────────────────────────────────────────────────────────────────────────────┤
│                                                                            │
│    ┌──────────┐         ┌──────────┐         ┌──────────┐                  │
│    │ Consumer │◄───────►│  Relay   │◄───────►│ Consumer │                  │
│    │  Node A  │         │  Node R  │         │  Node B  │                  │
│    └────┬─────┘         └────┬─────┘         └────┬─────┘                  │
│         │                    │                    │                        │
│         │    GossipSub       │    GossipSub       │                        │
│         │    (capacity)      │    (capacity)      │                        │
│         ▼                    ▼                    ▼                        │
│    ┌─────────────────────────────────────────────────────┐                 │
│    │              Kademlia DHT (Discovery)               │                 │
│    │         "Find nodes offering GPT-4o capacity"       │                 │
│    └─────────────────────────────────────────────────────┘                 │
│         │                    │                    │                        │
│         ▼                    ▼                    ▼                        │
│    ┌──────────┐         ┌──────────┐         ┌──────────┐                  │
│    │ Provider │         │ Provider │         │ Provider │                  │
│    │ (Ollama) │         │(API Key) │         │ (Claude) │                  │
│    └──────────┘         └──────────┘         └──────────┘                  │
│                                                                            │
│    Transport: libp2p + QUIC (NAT traversal via DCUtR hole punching)        │
│                                                                            │
└────────────────────────────────────────────────────────────────────────────┘

What It Does

TIM routes inference requests to available nodes across a decentralized mesh:

• Local models (Ollama, LMStudio, llama.cpp)
• API keys (OpenAI, Anthropic, Google)
• Subscription headroom (Claude Max, ChatGPT Pro)

If it can answer a Completions API request, it's a valid source.

Philosophy

We route inference requests. That's it.

• Client owns context, orchestration, tool execution, sandboxing
• Network owns routing, discovery, reputation
• Unit of work: single request/response pair
• Stateless at the network layer

Status

Specification phase. No implementation yet.

See specs/tim-overview.md for the full specification.

Planned Tech Stack

• Language: Rust
• Transport: libp2p + QUIC
• Discovery: Kademlia DHT + GossipSub
• API: OpenAI Completions API compatible