The Problem
Last-mile delivery is affected by traffic, demand spikes, and partner availability. Swiggy needed real-time visibility and predictive insights to improve on-time deliveries.
LOGISTICS · REAL-TIME ANALYTICS
Swiggy Delivery Analytics
Real-time analytics and intelligence platform for Swiggy to monitor deliveries, predict ETAs, and optimize operations across cities.
GitHub ↗Live Demo ↗
Stack
DATA ENGINEERING
PythonPySparkKafkaAirflowPostgreSQLRedis
ML / ANALYTICS
Scikit-learnXGBoostProphetGeoPandasFolium
BACKEND
FastAPIUvicornWebSockets
FRONTEND
ReactTypeScriptRechartsMapbox GL JS
DEPLOYMENT
DockerAWS (EC2, S3, RDS)NginxGitHub Actions
What I Built
An end-to-end analytics platform with real-time dashboards, ETA prediction, demand heatmaps, and delay analysis to help operations teams make faster decisions.
The Difference
Real-time streaming + ML-powered ETA prediction + geospatial intelligence in a single platform for proactive operational optimization.
Numbers That Matter
2.4L+orders analyzed daily
89.7%on-time delivery achieved
-18%avg delivery time reduced
25+cities covered in platform
12K+delivery partners tracked
Engineering Notes ✎
Real-time Pipeline
Kafka + PySpark Streaming process millions of events per day with low latency and high reliability.
ETA Prediction
Gradient Boosted Trees model using traffic, distance, weather and historical patterns for accurate ETA prediction.
Geospatial Intelligence
Heatmaps and zone analytics help identify high-demand areas and optimize partner allocation in real time.