Delhivery announced progress on an AI-driven location intelligence initiative developed in collaboration with NVIDIA, aimed at improving digital mapping accuracy for India’s diverse geography and non-standardised addressing systems.
Unveiled at the India AI Summit, the initiative focuses on building an India-first mapping layer powered by NVIDIA accelerated computing technologies, including NVIDIA CV-CUDA and NVIDIA Nemotron open models. The company is evaluating capabilities such as address disambiguation, contextual inference from unstructured descriptions, and predictive sequencing to optimise last-mile logistics operations across densely populated urban environments.
India’s addressing ecosystem often relies on landmarks, local languages and informal descriptions, creating challenges for traditional global mapping systems designed around structured coordinates. Delhivery said the initiative will leverage petabytes of proprietary logistics data accumulated over years of operations to develop more locally relevant geospatial intelligence.
The proposed platform aims to enhance route planning, improve delivery precision, and support digital services that depend on accurate location data. By combining its shipment dataset with NVIDIA’s AI infrastructure, the company intends to explore new ways of solving large-scale geospatial problems unique to India’s logistics and commerce ecosystem.
Delhivery noted that the long-term goal is to build a scalable digital infrastructure layer capable of supporting a range of applications requiring real-time, context-aware location intelligence across Bharat’s rapidly evolving digital economy.