Srikhetra Mohanty

Fault Detection Model (Predictive Maintenance)

• Generated pre-emptive alerts for manufacturing equipment failure for a leading aluminum producer
• Achieved successful removal (~80%) of deteriorated equipment leading to reduction in down time as well as increase in production efficacy
• Post prototyping, scaled up the ML pipeline across all production units, resulting in complete overhaul of the maintenance protocols

Production Unit Temperature Soft Sensor

• Created a temperature soft sensor, which predicts temperature (usually >900 deg C) inside a production unit
• The on floor operators/supervisors leveraged the temperature predictions which was ~80% times accurate within 5 deg C, to gauge the stability of the production unit
• Post prototyping, scaled up the ML pipeline across all production units of the organization, which equates to ~$1234 saved per annum in destroyed hard sensors

Warehouse Demand & Capacity Planning

• Generated granular, stable & less latent forecasts through custom designed ensemble framework
• Achieved ~90% accuracy across multiple time spectrums which resulted in optimal warehouse capacity planning

Publisher Network Analytics

• Designed a self-sustainable dashboard to track the overall performance of all major advertising KPIs for all advertising units (search bars) available for a leading search advertisement platform
• Performed churn analysis of customers on ad-hoc basis for partner
  publishers