All Projects
Machine LearningCase Study·93% Match

Car Price Prediction — MLflow + Kafka + Debezium

Complete MLOps system for predicting used-car prices (Ford dataset). Architecture: Streamlit UI → FastAPI backend → PostgreSQL → Debezium CDC → Kafka → ML Service → MLflow (experiment tracking + model registry) → MinIO (artifact storage). Docker Compose orchestrates all services including Zookeeper, Kafka UI, Adminer, and MinIO Console. scikit-learn models served via MLflow with Plotly charts in the Streamlit frontend.

PythonMLflowDockerKafkaPostgreSQLZookeeperMachine Learning
ProductionCDCDebezium + KafkaMLflowExperiment trackingMinIOArtifact storeStreamlitFrontend★ Featured
CDC
Debezium + Kafka
MLflow
Experiment tracking
MinIO
Artifact store
Streamlit
Frontend
🔴 The Problem
▸

Model retraining was manual and disconnected from database changes

▸

No artifact versioning or reproducible experiment history

✅ The Solution
▸

Debezium CDC captures every row change in PostgreSQL and streams it to Kafka

▸

MLflow tracks all experiments; MinIO stores model binaries with version metadata

📈 Impact & Results
▸

Any price data change triggers automatic downstream model refresh via Kafka

▸

Full reproducibility: every experiment is logged with parameters, metrics, and artifacts

▸

Streamlit UI with Plotly charts delivers predictions with feature importance

Full Tech Stack
PythonMLflowDockerKafkaPostgreSQLZookeeperMachine LearningMinIODebeziumStreamlit

More Projects

Agentic AI
Disaster Response AI Copilot
Machine Learning
Crypto Trade Prediction
Machine Learning
Heart Disease Prediction with Generative AI
Machine Learning
Movie Recommendation — Cosine Similarity & Collaborative Filtering
Interested in working together?
Let's build something impactful.