At Marsh, I own AI/ML operations across six production systems and improve workflows that support business and technical teams. I designed automated alerts for missing files and workflow failures, helping reduce project failures by ~70% and recurring failures by ~95%.
I build SOPs, runbooks, and training guides that make complex workflows repeatable. I’ve trained 30+ colleagues across seven applications and continue to monitor work quality after handoff.
Before my current role, I supported the production rollout of an ML prediction product for annual revenue planning, which achieved greater than 93.8% validation-year revenue accuracy. I also built forecasting models across five workforce drivers and six countries, achieving average MAPE under 9%.
