Technical Architecture

How Lurk Works

An arbitrage engine connecting decentralized prediction markets with institutional-grade analytics.

System_Diagram_v1.0
Client Layer
Next.js 15App Router / React 19
Edge API
/api/markets  |  /api/whales  |  /api/ai
Storage
PostgreSQLSupabase
Hot Cache
RedisRate Limiting / Queues
Ingestion Pipeline (Python)
Polymarket Fetch
Alchemy RPC / Polygon
Arb Calculator
Spread Logic / 1-3s Ticks
Kalshi Fetch
WebSocket + REST

Core Stack

Frontend
Next.js 15, Tailwind, Shadcn/UI, Recharts
Backend
Python FastAPI, Supabase, Redis, BullMQ
Blockchain
Alchemy RPC (Polygon), RainbowKit
Intelligence
Anthropic Claude 4.5 Sonnet

Data Sources

PolymarketCHAIN

Direct ingestion via Alchemy Polygon RPC nodes listening to Gnosis CTF contract events. Zero-latency state reading.

KalshiREGULATED

Institutional WebSocket feed combined with REST API polling for order book depth and spread calculation.

SentimentAI / SOCIAL

Custom scraper infrastructure feeding text data into Claude for sentiment scoring (-100 to +100).

SOCIALS

/ lərk / verb

to be concealed but capable of being discovered.

LURK AI • EST. 2026 • ATLANTA,GA

DATA DELAYED 15MS FOR STARTER TIER • INSTITUTIONAL GRADE