Skip to content
Adarsh Dwivedi.AI & Product Engineer
02Work

Adaptive Graph Search Suite

C++20 Routing Engine on Indian Road and Rail Networks

Adaptive Graph Search Suite running: C++20 Routing Engine on Indian Road and Rail Networks.

Captured from the live deployment, not a mockup

01 · The problem

Implemented 12 routing algorithms in C++20 over 1.7M junctions spanning Delhi NCR, 9 cities, the national highway grid and 70k rail nodes, from BFS and Dijkstra through to the techniques production engines actually use.

02 · How it works

  • 01

    Built Contraction Hierarchies reaching 44x faster with 460x fewer settled nodes than Dijkstra on the national highway network, plus Customizable CH that re-costs a 207k-node network in 38 ms instead of rebuilding in 1,222 ms.

03 · What it cost, and what it returned

Verified every route exact against a reference Dijkstra rather than trusting the speedup, added FIFO-safe time-dependent routing with a measured 2.02x rush-hour penalty, and compiled the engine to WebAssembly so it runs in a browser.