Data layer — torrent-like & private

Planned. The data layer is on the roadmap — this page describes the design, not a live system. See the Roadmap.

Deep bio & DNA data is huge (a single human genome is ~100 GB raw) and sensitive. Gennode will move and store it the way BitTorrent and IPFS move data — but encrypted, replicated and incentivized.

How it will work

• Content-addressed — every file and chunk will be identified by its hash, so it's tamper-proof and de-duplicated.
• Chunked — large files will be split into pieces (like BitTorrent), so many nodes share the load.
• Encrypted first — data will be encrypted client-side before it is sharded. Nodes store shards they cannot read.
• Replicated — erasure coding (Reed–Solomon) will keep data durable even when nodes go offline.
• Peer-to-peer — nodes will fetch pieces from many peers in parallel (a swarm); a DHT maps content → who holds it, with no central index.
• Access by key — only the data owner's keys can decrypt. Nothing is sold or profiled.

Why it matters

• Scale — spread terabytes of genomic data across thousands of nodes.
• Privacy — encrypted shards; operators never see raw data.
• Censorship-resistant — no single point of control or failure.
• Compute near data — jobs run where the (encrypted) data already lives.
• Model distribution — model weights, including Gennode's bio model, will be served the same way.

Earning (once the data layer ships)

A node will earn $GENNODE for storing and serving data shards — proven with storage proofs — on top of compute and uptime, once storage proofs ship. Today, points come from uptime × capacity; you'll be able to contribute GPU, disk, or both.

Think private, encrypted BitTorrent for biological data — with rewards for the people who keep it alive.