Madhushini Narayana Prasad

• Spearheaded cross-product collaborations with business partners to optimize costs and operational efficiency, resulting in multi-million dollar savings and significant operational efficiency gains.

Multi-echelon mix and blend optimization with end to end supply chain planning and inventory optimization for grains

• Goal: End to end supply chain optimization for grains from farms to final port of exit optimizing both storage and transport.
• Role: Lead optimization scientist and developer responsible for developing algorithms and implementing models to optimize profits accounting for (i) Inventory cost (ii) Quality and storage constraints (iii) Demand and transportation cost.
• Challenges: Problem was a multi-objective optimization problem with inherently complex business rules and constraints. Solution was required to work in real time with reasonable computational effort to enable what-if analysis and scalability to multiple locations.
• Impact: Deployed in South American countries and enabled significant multi-million dollar savings from increased throughputs, efficiency and lower wastage.

Shrimp farm health and yield optimization.

• Goal: Recommend feeding and harvest schedules to optimize shrimp growth and yield.
• Role: Identified and developed the business opportunity from conception and developed (i) prediction models to estimate the weekly growth and mortality rates of shrimps (ii) combined prediction modeling with optimization to optimize harvesting recommendations.
• Impact: Initial pilot deployed and showed significant yield benefits. Project expansion to multiple sites currently being scoped.

Texturizers blend prediction and optimization.

• Goal: (i) Reduce the lag time between proposing a blend to estimating its effectiveness (ii) Recommend blends based on end product specifications.
• Role: Identified opportunity, proposed to business leads and worked with business partners to create prototype. Lead data scientist and developer responsible for problem formulation, model development and end to end accuracy estimations.
• Impact: Initial prototype deployed in product segment was very successful and is currently being scaled to other plants that have large lead times from blend to effectiveness estimate.

• Worked on a variety of optimization projects both operational and strategic across business segments.

End to end inventory replenishment system based on demand profiles

• Role: Optimization tech lead responsible for developing models and algorithms to optimize sourcing, transport and inventory based on-demand profiles and supply constraints.
• Delivered end to end model by collaborating with key stakeholders across multiple domains including data-engineers, IT and business holders.
• Modeled and actively deployed the product across multiple geographies. Enabled smooth product deployment to end customers by actively analyzing deployment data and fine-tuning models and algorithms based on customer feedback.

Refinery Tankage Simulation Study

• Problem: Understand bottlenecks in refinery storage and identify optimal storage configurations and facilities for future demand and extreme conditions.
• Role: Technical lead responsible for developing the key solution and algorithms. Employed a combination of simulation techniques based on data to identify bottlenecks and optimization techniques to suggest configurations.
• Product was used to alter existing facilities resulting in better utilization and cost savings. Also used as primary vehicle for scenario analysis for strategic planning.

Other projects

• Extensively worked with other business partners such as pricing under uncertainty in demand / supply (market modeling and optimization), multi-product planning and optimization under shared facilities.

• Pricing and sales modeling: Developed models based on historic data to analyze the relationship between different factors that contribute to final sale price and quantity.

• Focused on the continuous refinement and improvement of the pairing optimizer software, which is one of the core algorithms in the airline crew-scheduling product used to generate flight pairings and assign crew types to flight schedules.
• Simplified the optimization model to reduce convergence time while maintaining solution quality.

• Provided strategic and tactical level network and transportation optimization solutions to Caterpillar business units from around the world.
• Example projects include facility location and sizing studies, shuttle route planning and inventory simulation.