Portfolio
Fleet Analytics & Driver Planning Platform
Logistics & Transportation
40%
Driver-vehicle balance visibility
100K+
Anomaly types detected automatically
15min
Reports generated daily
Overview
I built a fleet analytics platform for a logistics company managing 300+ trucks and 400+ drivers. The system aggregates data from multiple internal sources—scheduling system (Navigator), HR database, and vehicle registry—to provide unified reporting on driver-vehicle balance, anomaly detection, and operational metrics.
Business Context
A Polish logistics company managing 324 trucks and 470+ drivers was struggling with operational visibility. Their data was siloed across Navigator (scheduling system), HR database, and vehicle registry. Dispatchers, HR planners, and fleet managers each used different tools with no unified view. Planning decisions were made without complete information, leading to idle vehicles and driver shortages. The internal backoffice system needed integration for automated payroll deductions based on checklist compliance.
Challenge
Operations data scattered across scheduling system, HR database, and vehicle registry with no unified view of driver availability vs fleet capacity.
- Data scattered across scheduling system, HR database, and vehicle registry with inconsistent formats
- No visibility into driver-to-vehicle ratio and availability forecasting
- Manual detection of anomalies (sold vehicles still scheduled, drivers without employment dates)
Solution
We designed a unified data pipeline using Python and Pandas that aggregates daily operations data, calculates driver-vehicle balance metrics, and automatically detects 7 types of operational anomalies.
- Built Python data pipeline aggregating scheduling, HR, and vehicle data sources
- Developed 7-type anomaly detection system for operational issues
- Created integration module for internal backoffice payroll deductions
Approach & Methodology
We started by mapping data flows between systems and identifying key stakeholders: dispatchers, HR planners, driver planners, and workshop advisors. Using iterative feedback cycles, we built a Python-based data pipeline that aggregates daily CSV exports, applies temporal filtering for accurate historical analysis, and generates the reporting we needed. The system produces both simplified balance reports for executives and detailed anomaly reports for operations teams.
Implementation Details
Intelligent Aggregation Algorithm
Built an aggregation engine that tracks vehicle status (working, service, available) and driver status (working, vacation, sick leave), and calculates daily balance metrics with temporal filtering for employment/dismissal and purchase/sale dates.
Anomaly Detection System
Developed a multi-type anomaly detector identifying: drivers missing from HR database, sold vehicles still in schedule, HR data gaps (null employment dates), work-service conflicts, vehicles without tasks, and status inconsistencies.
Key Decisions
- Chose Python with Pandas over Go for data pipeline due to faster iteration speed and better data analysis capabilities—stakeholders needed frequent report format changes
- Implemented set-based operations for vehicle calculations to eliminate double-counting in work-service conflicts and ensure accurate availability metrics
- Built modular anomaly detection with configurable categories allowing operations to enable/disable specific checks as data quality improves
Tech Stack
Related Services
The following services were utilized in this project to deliver successful outcomes.
Lessons Learned
- Temporal filtering is critical—filtering by employment dates and vehicle purchase/sale dates eliminated major calculation errors in historical reports
- Data normalization from different systems is harder than expected—invested heavily in validation layers and anomaly detection to surface quality issues
- Simplified reports for executives proved more valuable than detailed ones—7-day balance summary with key metrics became the most-used feature
Project Information
Timeline
3 months for core pipeline, ongoing iterations
Team
1 developer + 1 analyst + stakeholder collaboration
Results
40%
Driver-vehicle balance visibility
100K+
Anomaly types detected automatically
15min
Reports generated daily
Have a Similar Challenge?
Let's discuss how I can help your project succeed with proven architecture and AI solutions.