All projects

Data & AI · 2026

Real-Time Tweet Sentiment Analysis

An end-to-end big data pipeline that trains a sentiment classifier on 1.6M tweets with PySpark MLlib and serves live predictions with Kafka and Spark Structured Streaming.

What I did

  • Trained a Logistic Regression model with PySpark MLlib on the Sentiment140 dataset (1.6M labeled tweets).
  • A Kafka producer streams tweets into a Kafka topic; Spark Structured Streaming classifies them in real time.
  • Predictions are stored in MongoDB and shown on a live dashboard through a Flask API.
  • The whole system runs with Docker Compose.

Tech stack

  • Python
  • PySpark
  • Spark Streaming
  • Apache Kafka
  • MongoDB
  • Flask
  • Docker Compose

Related projects