← Back to Portfolio
📈

AI-Powered Market Analysis
for Solana & Pump.fun Coins

Real-time machine learning model that predicts cryptocurrency price trends, analyses trading volumes, and identifies profitable opportunities across Solana and Pump.fun tokens.

Python TensorFlow Solana API AWS Node.js Machine Learning
View Source Code → Back to Portfolio

🎯 Project Overview

Built an end-to-end AI pipeline that ingests live cryptocurrency market data from multiple Solana-based exchanges and the Pump.fun launchpad, applies machine learning models for price prediction and volume analysis, and automates alerting for profitable trading opportunities. The system was designed to support informed retail investment decisions with real-time signals.

⚡ Key Features

📊 Real-Time Price Prediction

ML-powered forecasting of Solana and Pump.fun token prices using historical patterns, volume signals, and market momentum indicators.

🔗 Multi-Exchange Integration

Aggregates order-book and trade data from several crypto exchanges and news feeds to enhance prediction accuracy and reduce single-source bias.

🚨 Automated Alert System

Triggers real-time notifications when model confidence crosses profit thresholds — supports actionable investment decisions without manual chart-watching.

📈 Volume & Trend Analysis

Classifies trading volume regimes (accumulation, distribution, breakout) and overlays them with predicted price paths to surface high-probability setups.

📐 System Capabilities

Real-Time
Inference
Multi-Source
Data Ingest
Automated
Alerts
Cloud
Deployment

💻 Sample Model Output

Example prediction payload from the inference API:

# GET /api/predict?symbol=SOL&horizon=1h
{
  "symbol": "SOL/USD",
  "timestamp": "2026-04-22T10:15:00Z",
  "current_price": 172.48,
  "prediction_horizon": "1h",
  "predicted_price": 174.91,
  "expected_change_pct": +1.41,
  "confidence": 0.82,
  "volume_regime": "accumulation",
  "signal": "BUY",
  "sources": ["binance", "coinbase", "pumpfun", "news_sentiment"]
}

🏗️ Architecture & Stack

Backend & ML

Python ML pipeline training models on aggregated historical + streaming crypto data. TensorFlow for prediction models, Node.js orchestration.

PythonTensorFlowPandasNode.js

Data Sources

Integrates with Solana JSON-RPC, Pump.fun APIs, and centralized-exchange REST endpoints for multi-venue market data.

Solana APIPump.funExchange RESTNews Feeds

Deployment

Containerized deployment on cloud infrastructure with scalable inference endpoints and scheduled retraining jobs.

AWSRenderDockerCI/CD

Frontend Dashboard

Interactive dashboard for visualising predictions, historical performance, and alerts, built with Chart.js and vanilla JS.

HTMLCSSJavaScriptChart.js

🛣️ Roadmap & Future Enhancements

  • 🔒 On-chain wallet integration for automated trade execution (Solana Web3.js)
  • 🧠 Transformer-based sentiment analysis on crypto Twitter and news articles
  • 📱 Mobile push notifications for alerts (iOS / Android via FCM)
  • 📊 Back-testing framework with PnL simulation across historical periods
  • 🧪 Reinforcement-learning trading agent for auto-rebalancing strategies

Interested in the code?

Full source, ML training scripts, and dashboard frontend are on GitLab.

View Source on GitLab →