🔗 LinkedIn пост · Gayathri G Murali · AI & ML Engineer | GenAI & NLP | Geospatial AI | Agentic Workflows | Designing Intelligent & Scalable Pipelines · 2026-05-27 · 👍 8 · 💬 2 · 🔁 0
🚀 Building a Hyperlocal Dispatch Optimization Engine using Real Google Maps & Folium Data 🛵💨
Hyperlocal delivery logistics are all about fighting against time and spatial constraints. To closely understand how quick-commerce giants manage multi-drop dispatches efficiently, I built an algorithmic Order Batching Optimization Engine mapped specifically for the Oachira grid in Kollam, Kerala — a region I know inside out.
Instead of relying on synthetic coordinates, I extracted real-world location data directly from Google Maps to make this simulation as close to reality as possible.
📌 The Real-World Bottleneck:
The standard “one-rider-per-order” dispatch model creates a massive operational overhead. When multiple riders traverse overlapping paths from the same restaurant hubs to nearby drop locations, it leads to sub-optimal fleet utilization, increased driver fatigue, and unnecessary fuel burn.
🛠️ Architecture & Engineering:
Geospatial Hotspot Mapping: Ingested real-world coordinate data for 10 accurate food hotspots inside Oachira (extracted via Google Maps) and simulated concurrent, real-time demand loops for 5 active customer delivery slots.
Spatial Routing Engine: Implemented the Haversine Formula to compute precise geodesic point-to-point distances over the dynamic grid to forecast realistic ETAs based on local speed constraints.
Dynamic Batching Heuristic: Built a greedy clustering mechanism with a 1.5 km proximity constraint to dynamically pool concurrent orders…