Arbitrage operator

Find the gap. Scale until it closes.

I look for gaps in infrastructure and markets, then build the systems to work on them. I operate Web Data Infrastructure (WDI), deploying and maintaining distributed nodes for public-web data access, alongside my exchange research.

Explore WDI ↗ Get in touch

Around the internet as

cosmosgeek

@cosmosgeeketh

THE JOURNEY

2026—2022.

Current work first.
Scroll back through the years.

NowWDI
2025–26Trader
2025Cross-chain
2023–25Airdrops
2023Markets
2022–23Mining
2022Bandwidth

NOW / OPERATING INFRASTRUCTURE

Web Data Infrastructure ↗

Public-web infrastructure.
Built, operated, measured.

I operate a distributed network of containerised applications for public-web data access—from deployment and runtime health to traffic recording and recovery.

Explore Web Data Infrastructure ↗

Traffic overview

Connecting

Network illustration · drag to rotate

30-day traffic: —MB
24-hour traffic: —MB
Total unique IPs recorded: —
Current rate: —MB/min

Awaiting readings

MB/min · 1S readings

Live network traffic · received + sent bytes

06 / 2025–2026

Trader & exchange systems

Binance · Python · Tokyo

From watching markets to writing the system that reads them. I built market scanners, basis research and an execution program—working through data arrival, trade economics, paired orders and position management.

Activity against its own baseline.

One-minute quote volume from a saved market scan.

MOG: 119.11× its median minute. PEPE: more volume, but only 1.66×.13 Aug 2026 · 23:19 UTC

TRADER / BINANCEARCHITECTURE · EXPLORE THE PATH

market.py

Read the executable market.

Spot and perpetual order books, available depth and volume changes. Prices you can trade against matter more than a headline spread.

WebSocket feeds→order-book depth→market scan

Normal basis · reverse basis · volume researchTrader · private repository ↗

Cosmos Arbitrage

Discovery and replay: executable basis and persistent volume changes across 1s, 5s and 15s windows.

Quant

Trade economics and lifecycle: depth, carry, liabilities, partial fills and recovery.

Trader

Binance execution: paired order handling, fill reconciliation and position exits.

05 / 2025 · CROSS-CHAIN AUTOMATION

Cross-chain scripts

Python · Final.py

In two months, I went from no programming background to cross-chain automation—iterating with ChatGPT to keep transactions moving, recover stalls and scale activity.

~$100Mcumulative cross-chain transaction volume
Built through two months of iteration.

Final.py / oUSDT ROUTING7 NETWORKS · 10 WALLET POSITIONS · ILLUSTRATED CYCLE

Base

Optimism

Unichain

Soneium

Ink

Sonic

Mantle

Source wallet Wᵢbridge →Destination Bᵢtransfer →Relay W₀transfer →Next wallet Wᵢ₊₁

Balance scan → allowance → gas quote → KMS sign → send → recover / continueHow Hyperlane routes tokens ↗

Final.pyConstruct

390 / next_nonce

with _nonce_lock:
    n = _nonce_map[key]
    _nonce_map[key] = n + 1
    return n

AWS KMS / EIP-1559 SIGNING

raw = kms_sign_eip1559(
    tx_try,
    alias
)

smart_send / SUBMISSION

txh = rpc(
    w3.eth.send_raw_transaction,
    kms_sign_eip1559(tx_try, alias)
)

smart_send / CONFIRMATION

wait_for_confirmation(
    w3, txh,
    required_confs=1,
    timeout=confirm_timeout
)

Source-chain receiptnext transaction ↗

Selected logic from the transaction workflow

Two months of iteration.
Every stalled transaction mattered.

I started without a programming background and iterated with ChatGPT until the workflow could move through wallets and chains, handle nonce conflicts and recover stalled sends.

Final.py

01ConstructRead balances, allocate nonce and quote cross-chain gas.
02SignECDSA signature, canonical low-s and sender recovery.
03SubmitRLP encoding and submission to the source RPC.
04ConfirmConfirm; retry, replace or clear a stalled nonce.

04 / LATE 2023–2025

Airdrops

Research → participation → scale

After learning the markets, I went deep into protocols: finding opportunities early, participating, tracking eligibility and scaling the work across accounts. I worked with @CC2Ventures. This time, the effort paid off.

Arbitrum was the opportunity I missed. Hyperliquid became one of the cycle’s defining distributions. My biggest result came through a different lane: Hyperlane.

Hyperlane

HYPERLANE / PERSONAL DISTRIBUTIONSBoth seasons$440,000USD received across two seasons

From smaller distributions to an operation built around cross-chain activity. Hyperlane rewards use of its messaging network.

Season 1 · 2024–25$25,000

Season 2 · summer 2025$415,000

See the code behind the activity ↗

2023 / THE CYCLE OPENS

Arbitrum

I missed this drop.

2024 / THE INDUSTRY MOMENT

Hyperliquid

A landmark community distribution.

2025 / MY BREAKTHROUGH

Hyperlane

Two seasons. A much larger result.

Across other protocols

Personal distributions · USD

ZKsync$15,000
LayerZero$8,000
EigenLayer≈ $3,000
Scroll$2,000
Celestia≈ $1,500
MANTRA Chain≈ $700
deBridge$500
Jupiter≈ $300

Hyperlane’s distribution and rewards model ↗

03 / 2023

Learning the market from the inside.

Binance spot, then futures, then options. I learned from scratch—watching markets and order books until I could trade profitably.

Studying liquidity, execution and risk laid the groundwork for the trading systems I later built.

BTC / USDT BINANCE2023 · REAL WEEKLY CANDLES

Loading the 2023 market…

Historical market data, not my account performanceBinance public market archive ↗

02 / THE FOLLOWING MONTHS

Making mining work.

Late 2022–2023 · Kaspa · Linux · Docker

I spent the next few months writing code, packaging mining applications and trying to scale profitable runs. Getting a miner to start was only the beginning.

The real test was whether the rewards could cover running costs. After months of development and iteration, this attempt also did not become profitable.

KASPA / MINING OPERATIONSSET UP · PACKAGE · RUN · REPEAT

A running miner.
A harder equation.

NVIDIA GPU product render illustrating the hardware behind mining

GPU illustration · NVIDIA ↗

Docker deployment / example commands

docker pull ubuntu:22.04

# Install the mining runtime
# Configure pool and worker settings

Application + environment

Mining rewards−Running costs=Not profitable

Months of code and scaling attempts · outcome, not simulated earningsWhat determines mining profitability ↗

01 / LATE 2022

The first scaling problem.

Bandwidth applications

I started by running bandwidth-sharing apps. Then the question became: how do I put the same apps into one repeatable setup, keep them alive, and scale that setup towards a thousand containers?

Months of setup, failures and iteration. The idea was to make scale pay. The economics did not work out—but the infrastructure work started here.

APPLICATIONS → ONE SETUP → A FLEETINTERACTIVE SCALE MODEL

One repeatable setup.

Peer2Profit
Repocket
Honeygain
PacketStream
EarnApp
ProxyRack

Package the applications.
Repeat the environment.

1containers in the model

Explore the scale ↗

Scale target: 1,000 containers · outcome: not profitableReal application / Docker image ↗

Stack

The tools behind the projects

Python
Docker
Linux
AWS
AWS KMS
Cloudflare
Git
SQLite
JavaScript
TypeScript
HTML
CSS
WebSockets
Asyncio
Web3.py
C / C++

HAVE SOMETHING WORTH BUILDING?

Let’s worktogether!

Contact details ↓

Find the gap. Scale until it closes.

2026—2022.

Web Data Infrastructure ↗

Traffic overview

Trader & exchange systems

Activity against its own baseline.

The executable gap.

Measure each part of the path.

Read the executable market.

Cosmos Arbitrage

Quant

Trader

Cross-chain scripts

Airdrops

Arbitrum

Hyperliquid

Hyperlane

Across other protocols

Learning the market from the inside.

Making mining work.

The first scaling problem.

One repeatable setup.

Stack

Let’s worktogether!

There’s no résumé.

Let’s talk.

Find the gap. Scale until it closes.Find the gap.Scale untilit closes.it closes.

2026—2022.

Traffic overview

Activity against its own baseline.

The executable gap.

Measure each part of the path.

Read the executable market.

Cosmos Arbitrage

Quant

Trader

Arbitrum

Hyperliquid

Hyperlane

Across other protocols

One repeatable setup.

Stack

Let’s worktogether!

Find the gap. Scale until it closes.