Mac Mini Scan
A real-time monitoring and signal-detection service in Python. Polls Apple's per-store fulfillment surface nationwide, normalizes it into a clean availability model, and fires an alert the instant a configuration flips to in-stock. Acted on that signal to source and resell just under a dozen units for close to $10,000 in a single month.
Overview
A real-time monitoring service that detects a state change across a few hundred distributed endpoints and turns it into an actionable alert faster than a human could poll. The system continuously polls Apple's per-store fulfillment surface for every Apple Store in the US, normalizes an undocumented and unstable response into a clean per-model / per-store availability model, and fires the instant a configuration flips from unavailable to in-stock anywhere in the country. Built in Python around three constraints: the upstream is not a stable public contract and its response shape can change without notice, so parsing is defensive and degrades gracefully rather than crashing; the polling loop has to be rate-aware and polite enough to run continuously for weeks without being blocked; and the entire value is latency, because a signal that arrives late is worth nothing. The problem it was built for was the Mac Mini and Mac Studio shortage driven by surging RAM demand and the rush to run AI models locally, where high-memory configurations sell out in minutes and reappear unpredictably. The alerting proved out commercially: acting on the signal sourced and resold just under a dozen units for close to $10,000 in a single month. The underlying pattern generalizes to any monitoring problem where an event has to be detected, normalized, and acted on in near real time. Run privately. Source is private.
Features
- Nationwide coverage — every Apple Store checked, not a single region
- Real-time stock-flip detection (unavailable → in-stock) for Mac Mini + Mac Studio
- Per-model / per-store availability map, normalized from Apple's fulfillment data
- Latency-first design — surfaces a buyable unit before a manual refresh would
- Rate-aware, polite polling loop that holds up over long continuous runs
- Resilient parsing that tolerates upstream response-shape changes
- Instant restock alerting that turns a drop into an actionable buy signal
Technologies Used
- Python
- Continuous web scraping / polling of Apple's per-store fulfillment surface
- Real-time monitoring + change detection
- Response normalization into a per-model / per-store availability model
- Alerting pipeline for instant restock signals
Challenges & Solutions
Challenges
- High-memory Mac Mini / Mac Studio configs sell out in minutes during the RAM-driven shortage
- Restocks appear unpredictably and at any of hundreds of stores nationwide
- Apple's fulfillment response shape is not a stable public contract and can shift
- The entire edge is latency — a late signal is a useless signal
Solutions
- Continuous nationwide polling so a restock anywhere is caught the moment it lands
- Defensive, tolerant parsing so upstream changes degrade gracefully instead of breaking
- Normalized availability model + instant alerts so a drop becomes an immediate buy decision
- Rate-aware loop that sustains real-time coverage without hammering the source