MasteryGPS
Real-time fleet tracking
Part-time freelance for Mastery Telematics: real-time tracking for 500+ devices, a resumable Python crawler over a public Moroccan company registry, and a Prometheus uptime stack that moved incident detection off customer reports.
Highlights
- 500+ devices monitored with 99.9% uptime
- Python scraping — 85% reduction in manual data entry
- Server monitoring with 95% early incident detection
Role
Part-time full-stack freelancer: owned the data-collection pipeline and uptime monitoring alongside the React + Node tracking surface used for 500+ machines.
Approach
Data collection ran as a resumable crawler over a public Moroccan company registry, segmented by the registry's 15 activity sectors and checkpointed to a log so a multi-day run could restart from its last position rather than the beginning. Throughput work drove the design through successive concurrency models — threads, then process pools, then an asyncio/aiohttp pipeline — with Selenium reserved for the client-rendered pages that plain HTTP could not reach. Output landed as both CSV and JSON so operations staff could open it directly.
Reliability
Uptime monitoring started as a Python poller with email and SMS escalation, then moved to a containerised Prometheus stack: Blackbox exporter probing the fleet's public endpoints on a 15-second interval, with Alertmanager routing notifications. That shifted incident detection from customer reports to automated alerts.
Outcome
React + Node tracking for 500+ devices at 99.9% uptime; the crawler cut manual data entry by ~85%; monitoring improved early incident detection to ~95%.