Sand production remains a key technical challenge in many reservoirs where formations comprise of weakly consolidated sandstone. Sand control completion equipment is typically installed to prevent sand from entering the well. However, in cases where the sand control is ineffective due to installation flaws/defects, high sand production may occur often requiring choking back of wells and resulting in significant hydrocarbon production losses. An effective remediation requires identification of locations of sand entry. However, there is currently no proven technology available in the market that accurately identifies downhole sand ingress locations in real-time. In this paper, we present results from a new technology solution that addresses this challenge by using in-well conveyed fibre optic distributed acoustic sensors (DAS) for the detection of sand ingress zones across the reservoir section throughout the production period in real time. The technology employs a novel signal processing technique that isolates and extracts acoustic signals resulting from sand ingress from background flow and instrumentation noise in real time. The new processing architecture also addresses the "big-data problem" that currently hinders DAS technology uptake through use of intelligent feature-extraction techniques that reduce data volumes at source (by several orders of magnitude). The technology feasibility has now been verified both through flow loop experiments and through multiple field trials and has been successfully used to inform the first targeted sand remediation in a BP production well.