AERIS
AERIS
Aeris samples network conditions continuously and reports them as they are — not as a forecast, not as a snapshot, not as a story someone wants told.
{ Instruments }
Every instrument reports a single reading, timestamped, with its margin of error attached. Nothing here is smoothed for comfort.
Live from Bridge & Client Watch
{ Bridge & Client Watch }
Aeris tracks ~180 IBC light clients across every chain connected to Celestia, catching risk-state transitions — healthy to expired, expired to recovered — as they happen, not on a weekly crawl.
No sampling window, no cached snapshot pretending to be live. If a client's state changes, the next check sees it — and so does the Signal Feed.
Open Bridge & Client Watch{ Signal Feed }
One feed instead of three tabs: network upgrades, governance proposals and bridge-health changes, ranked by severity with a public Atom feed for anyone who'd rather not check a dashboard at all.
Every entry traces back to the raw record that produced it — a vote, a client-state change, a proposal going live — never a summary written after the fact.
Open the Signal FeedLive from the Signal Feed
A measurement shop first, a validator second.Our public tools read chain state directly instead of repeating someone else's summary.
{ Recent Signals }
Pulled straight from the Signal Feed — not a curated highlight reel, just the most recent entries in order.
{ Standing Orders }
These hold regardless of market conditions, headcount, or pressure to ship faster.
§01
Assets remain in verifiable custody at every step. Nothing is rehypothecated without disclosure.
§02
We read state continuously, not on a schedule. No cached snapshot is ever presented as live.
§03
Every measurement carries a timestamp and a margin of error. Stale data is marked, never hidden.
§04
Anomalies are surfaced once, not repeated as alerts. We report weather, not forecasts.
§05
We state direction and uncertainty together. Certainty we don't have is never implied.
§06
Every published reading is traceable to the raw sample that produced it, indefinitely.
{ Duty orders }
Day watch
Bangkok
Watch the alerts, work the ticket queue, roll out upgrades and answer delegators.
Night watch
Hold the same duties through the night. The two operations engineers split the clock, so the nodes always have someone awake.
Note
Core routers, line cards and optics, traffic engineering and port upgrades planned months ahead. Applied here: choose the NICs and uplinks, and read packet loss and interface errors before they turn into missed blocks.
Instruments
Build and maintain the instruments at the top of this page: Bridge & Client Watch and the Signal Feed.
Network &
governance
Own provider and facility choices. Read every proposal on the networks we validate, prepare our votes and keep the hardware wallet. Came to this from transit and peering work.
All four
Commission and new-network decisions are made by all four together.
{ Who we are }
Four network engineers in Bangkok who came to validation from Thai mobile and carrier networks.
Two of us spent years at dtac (now part of True Corporation) in the operations centre and on the internet gateway: IP/MPLS backbone, mobile backhaul, BGP transit and peering, DDoS scrubbing and capacity planning for a network that carried millions of subscribers and could not go dark.
That work leaves one habit: a reading is only useful with a timestamp, a threshold and an honest margin of error, and a green dashboard never meant a healthy network to us.
LoggedEntry
Before
Backbone, backhaul, transit and peering for millions of subscribers.
Nov 2024
Three of us went to the node and infrastructure sessions out of curiosity.
Same week
Running it felt like the work we already knew: uptime, latency, disks, upgrades on a deadline.
Early 2025
Running in production on the networks listed below.
{ Networks }
The infrastructure runs where readings are needed. Each validator below carries live custody and is independently verifiable on its native explorer.
ZetaChain
zetachain mainnet
A universal L1 that observes Bitcoin, Ethereum and Cosmos natively, settling cross-chain state without wrapped assets or custodial relayers.
Archway
archway-1
A CosmWasm smart-contract chain that routes a share of inflation and gas back to the applications generating that activity.
Band
laozi-mainnet
Cross-chain oracle network providing verifiable data feeds to decentralized applications.
Terra
phoenix-1
Cosmos-based network for decentralized applications and assets.
{ Where we run }
For Celestia Mocha, the node and the Fibre machine run on dedicated hardware at CSL's CyberWorld data center in Bangkok, on the AIS network. Each network's signer sits at True IDC's East Bangna campus across the city.
Site A · Celestia Mocha
Data center, Bangkok
Site B · Every network
Campus, Bangkok
The signing path and the keys are never in the same building.
{ Why Celestia and Fibre }
Fibre turns data availability into the kind of problem we spent our careers on. A Fibre server has to take uploads from many clients at once, hold its share of rows on fast disks, keep latency to the validator's signer low and grow its link before the traffic does.
Those are the same questions we answered at the gateway every quarter: how much burst a port can take, where the bottleneck moves when throughput doubles, what fails first under load.
We readBecause