Flood Prediction System (Group)

🌊 Flood Monitoring – Data Intelligence

A FastAPI-based Data Intelligence microservice developed for a microservices-driven flood monitoring platform. My primary contribution was designing and implementing an end-to-end machine learning pipeline** using XGBoost to transform live environmental sensor data into real-time flood risk predictions.

🤖 End-to-End ML Pipeline • Designed and implemented a complete machine learning pipeline for flood risk prediction.
• Built and optimized XGBoost models using historical and real-time environmental sensor data.
• Implemented the full ML lifecycle including data preprocessing, feature engineering, model training, evaluation, hyperparameter tuning, inference, and deployment**.
• Generated low-latency flood risk predictions that integrate seamlessly with the platform's backend services.

⚡ Real-Time Intelligence • Delivered real-time flood risk predictions for early warning and decision support.
• Designed the inference pipeline for seamless integration within a microservices architecture.
• Produced prediction outputs consumable by dashboards and downstream services.

🗄️ Data Infrastructure • Utilized PostgreSQL + PostGIS for relational and geospatial data management.
• Leveraged InfluxDB for efficient time-series sensor storage.
• Integrated Kafka (aiokafka & confluent-kafka) and MQTT for reliable real-time data ingestion.

🛠️ Tech Stack Python, FastAPI, XGBoost, Scikit-learn, Pandas, NumPy, PostgreSQL, PostGIS, InfluxDB, Apache Kafka, aiokafka, confluent-kafka, MQTT, Socket.IO, Kong API Gateway, Clerk, JWT, Git, GitHub, OpenAPI**