Rishikesh K.

Road Truck Scheduler: a) Forecasting Sales--i) Implemented different forecasting algorithms like ARIMA, ARIMAX, Facebook Prophet, LSTM, Ratio Model, etc. to forecast sales of 3 different products for 100+ petrol stations using the past 3 years data. ii) improved accuracy by 2.3 % baseline (average method) b) Scheduler: Implemented three different methodologies to solve scheduling problem i) Heuristic with simulated annealing ii) Route generations with iterative MILP and iii) MILP for Multi-trip constraint Vehicle routing problem with time window having a unique character like a clustered customer, 3 major products and 5 major truck sizes, improving scheduling with modifying orders and re-scheduling orders. •TerminalOptimization: a) Forecasting Production: i) Implemented a LSTM model to forecast production from the past 5 years of production data. ii) improved forecasting by 2.5% and forecasted for 3 months ahead. b) Schedule Generator: i) implemented MILP formulation for scheduling vessel and pipeline crude oil transportation from the terminal to different partners and PETRONAS refinery. ii) Implemented schedule modification due to production variation while partners requirements. c) Scaling up the models to other terminals on DataBricks using PySpark and python. •VesselScheduling Problem: Supervised the team to work with TCS on inventory constrained vessel routing problem, found optimal CPLEX parameter to reduce runtime of the model. •LNG Annual Delivery Plan: Part of the team to formulate MILP for the annual calendar for vessel scheduling. •Lead team to build AzureML studio models to solve finance team projecting budget and profit projecting and PowerBi dashboard. •Workingon an Image analytics project to detect corrosion site images using deep learning algorithms (CNN).