Senior Research Scientist · Amazon

Science for Decision-Making at Scale

Optimization, forecasting, and AI for complex interconnected systems.

I build systems that make high-stakes decisions under uncertainty — routing, demand planning, capacity allocation, and multi-agent coordination. The common thread is combining mathematical optimization and machine learning to solve planning problems where decisions are coupled, stochastic, and time-sensitive.

At Amazon, this powers the optimization and forecasting behind last-mile delivery across a global network. Previously, I led operations research at BNSF Railway, contributing to work recognized with the INFORMS Prize (2018). More recently, my research has expanded into agentic AI systems and coordination in multi-agent environments.

What I work on →

PhD Supply Chain and Operations Management, UT Austin · 2 U.S. patents granted, 8 pending · Published in Production and Operations Management

Chinmoy Mohapatra
2 Granted · 8 Pending U.S. Patents
Global-Scale Logistics Impact
30+ Invited & Conference Talks

Recent Highlights

A snapshot of recent talks, patents, publications, and recognition

2026 Talk

CSS 2026 — Oral (OR & Optimization) + Poster (Machine Learning)

2026 Patent

Patent Filed — Multi-agent coordination systems

2025 Patent

2 USPTO Patents Granted — Consensus Planning & Storage Pod Design

2024 Paper

Published in Production and Operations Management — Subscription Pricing

2024 Talk

Invited Panelist & Tutorial Speaker — INFORMS, POMS

2022 Award

Top 2 OR Paper Award — Amazon Machine Learning Conference

Systems That Ship

Production systems I've built or led — deployed science at scale

Package Selection Systems

Deployed across Amazon's US delivery network

A family of optimization systems that determine which packages to assign to which delivery stations and routes, maximizing network density while balancing cost, speed, and capacity constraints across the entire fulfillment-to-doorstep pipeline.

Millions of packages daily across 1,000+ stations

Demand Forecasting Systems

6 regions, 4 time horizons, 10M+ weekly forecasts

A large-scale forecasting platform managing concurrent prediction models across multiple global regions, combining classical statistical, deep learning, and tree-based ensemble methods with hierarchical reconciliation to drive capacity and staffing decisions.

1M+ concurrent models across 6 global regions

Capacity Planning & Supply Chain Coordination

End-to-end supply chain, multiple planning horizons

Systems that allocate delivery capacity under demand uncertainty and coordinate across supply chain stakeholders — from warehouse labor planning to network-wide resource balancing — ensuring service targets are met even as conditions shift.

Multi-region network coordination under demand uncertainty

Selected Research

Publications, patents, and workshop papers with authoritative links

Paper
Subscription Pricing for Free Delivery Services

Production and Operations Management, 33(4), 943–961, 2024

Patent
Designing Storage Pods with Layers of Bins or Slots

US 12,504,281 B1 · Granted 2025 · Google Patents

Speaking

I speak on demand forecasting at scale, last-mile logistics optimization, the intersection of OR and ML in production systems, and the path from research to deployed decision systems.

Demand Forecasting at ScaleLast Mile Logistics OptimizationBridging OR & ML in Production Systems
Explore Speaking Topics & Past Appearances

Research, Speaking & Collaboration

Whether it's a research collaboration, a speaking invitation, or just a good conversation about OR and AI — I'd welcome the connection.

Get in Touch