The problem
Crypto markets run continuously, which makes manual strategy testing and execution inconsistent. I wanted a system that could collect exchange data, transform it quickly, and execute algorithmic trading ideas through a repeatable Python workflow.
What I built
I built an automated trading system around the Binance API with real-time market data ingestion, Python-based feature engineering, and ML-informed strategy logic. The project used Pandas and NumPy for research workflows, Numba to speed up numerical loops, and AWS for deployment.
- Connected Binance API market data to a Python processing loop.
- Implemented algorithmic trading strategies using Pandas, NumPy, and Numba.
- Built a deployment path on AWS so the system could run outside a local notebook.
- Used the project to practice production habits around automation, monitoring, and repeatable execution.
24/7
market monitoring window
6
core technologies across research and deployment
Technical shape
def score_signal(features):
momentum = features["close"].pct_change().rolling(12).mean()
volatility = features["close"].pct_change().rolling(24).std()
return momentum / volatility.clip(lower=1e-6)
What changed
This project became the bridge between applied mathematics, machine learning, and production-minded engineering. It gave me hands-on practice with real-time data, automated decision loops, API reliability, and the discipline required when model output connects to live execution.