Talk Abstract

Most mining operations are sitting on an untapped asset: decades of historical drilling, logging, and mapping data. While traditional geostatistical methods remain a cornerstone of resource estimation, new advancements in deep learning allow us to look at this data through a different lens. By training neural networks to recognize complex spatial patterns across multiple data types, including qualitative geological logs, we can identify subtle indicators of mineralization that might not be captured by standard estimation alone.

In this talk, we will discuss how these predictive models are being applied to near-mine exploration and brownfield expansion. Using examples from real case studies, we will show how moving toward these data-driven methods can significantly improve drill targeting and resource confidence. The goal is to demonstrate how geologists can use the information they already have to uncover new ore zones and maximize the value of their existing assets.

Speaker Bio

Leo Iakovlev, P.Geo, is Chief Resource Geologist at Stratum AI with nearly two decades in the mining industry, including over a decade in resource estimation and technical evaluation. He has worked across major mining companies, consulting, on projects ranging from greenfield exploration to development and M&A. At Stratum AI, he applies geological expertise with AI-driven methods to improve data quality, uncertainty management, and decision-making in mineral resource evaluation.