In a novel pilot program, the Metropolitan Transportation Authority (MTA) partnered with Google Public Sector to explore the potential of artificial intelligence in revolutionizing subway track maintenance. We utilized onboard sensors in low-cost Google Pixel phones to capture data on track conditions while mounted to running subway cars. This matured to an AI and Machine Learning solution with the ability to proactively identify and address potential track nonconformities, often through audio alone. This initiative marks a significant step towards a more efficient, reliable, and safer subway system, showcasing the power of AI-driven predictive maintenance in critical infrastructure. When paired with the right user experience for track inspectors to harness the power of these insights, we can shine a light into tunnels of otherwise invisible risk.