ML / Data ScienceProtected2026

Sol-Sniper

AI Solana trading bot (MCP)

A walk-forward-validated LightGBM pump-probability model (AUC 0.82) feeds a staged Claude-vision agent pipeline with on-chain execution and full risk guardrails, packaged as a 14-tool MCP server running in shadow (paper) mode.

ML / Data Science · PreviewSol-SniperAI Solana trading bot (MCP)

Local MCP server with a live dashboard; trades run in shadow (paper) mode.

Protected work

A live trading system, so execution strategy, RPC keys, and on-chain wiring are kept private, and it runs in shadow (paper) mode. Glad to walk through the ML and architecture.

Request access

Sol-Sniper discovers tokens (Telegram signals, Pump.fun graduations, a Playwright DexScreener scraper), scores each candidate through a deterministic scorer and a walk-forward-validated LightGBM pump-probability model, then runs a staged Claude-powered agent pipeline (Scout → Chart Analyzer → Analyst → Portfolio) where chart screenshots are sent to the Claude vision API for pattern analysis. Trades execute on-chain via Jupiter and Pump.fun with a shadow (paper) mode, hardened with a circuit breaker, policy guardrails, preflight gates, RPC key rotation, and a backtest engine. Packaged as a local MCP server with a FastAPI + WebSocket dashboard.

Python
LightGBM
scikit-learn
MCP
Claude API (vision)
FastAPI
WebSockets
Playwright
Solana (solders)
Jupiter Swap API

Architecture · discovery → ML scoring → agent decision → guarded execution

Token discovery
Telegram signals, Pump.fun graduations, and a Playwright DexScreener scraper surface candidate tokens.
ML scoring
A deterministic scorer + a walk-forward-validated LightGBM pump-probability model (AUC 0.82) rank each candidate.
Claude agent pipeline
A staged pipeline (Scout → Chart Analyzer → Analyst → Portfolio) sends chart screenshots to the Claude vision API for pattern analysis.
Risk guardrails
A circuit breaker, policy guardrails, preflight gates, RPC key rotation, and a backtest engine gate every decision.
On-chain execution
Trades execute via Jupiter / Pump.fun, in shadow (paper) mode by default.
MCP server
Packaged as a local MCP server (14 tools) with a FastAPI + WebSocket dashboard.

Pump model AUC: 0.82 (walk-forward)
Training samples: 7,188
MCP tools: 14
App code / tests: ~16.5K LOC · 1,072 tests

Private · request access

Want something like this? Get in touch →

Architecture · discovery → ML scoring → agent decision → guarded execution

Token discovery

Telegram signals, Pump.fun graduations, and a Playwright DexScreener scraper surface candidate tokens.

ML scoring

A deterministic scorer + a walk-forward-validated LightGBM pump-probability model (AUC 0.82) rank each candidate.

Claude agent pipeline

A staged pipeline (Scout → Chart Analyzer → Analyst → Portfolio) sends chart screenshots to the Claude vision API for pattern analysis.

Risk guardrails

A circuit breaker, policy guardrails, preflight gates, RPC key rotation, and a backtest engine gate every decision.

On-chain execution

Trades execute via Jupiter / Pump.fun, in shadow (paper) mode by default.

MCP server

Packaged as a local MCP server (14 tools) with a FastAPI + WebSocket dashboard.