BigAl Sniper Bot
A low-latency Solana trading bot engineered for sub-200ms transaction construction. AWS EC2 with CPU pinning, Rust transaction builder, Jito bundle submission, and a real-time terminal UI.
What we set out to solve.
On-chain memecoin launches resolve in milliseconds. Off-the-shelf trading clients add network hops, garbage-collected runtimes, and shared-tenant CPU contention that put any retail bot a tick behind. Latency is the entire alpha.
How we built it.
Hot path in Rust
The transaction builder is written in Rust with statically allocated buffers and zero heap churn on the hot path. Signing keys live in memory; instruction layouts are precomputed at startup.
CPU pinning and EC2 placement
The bot runs on an EC2 instance in us-east-1 (Jito region), with the executor pinned to a dedicated physical core via taskset. The Python orchestration layer runs on separate cores so it cannot preempt the hot path.
Jito bundling
Submitted transactions ship as Jito bundles with tip configuration tuned per launch profile. WebSocket dashboard streams every signed bundle, slot landing, and PnL in real time to a Fidelity-inspired terminal UI.
The numbers.
What it changed.
A vertically integrated trading stack — from market data ingest through signed transaction land — that competes on the same latency floor as professional desk infrastructure, controlled from a single browser-based dashboard.