Sumedha Katarya, Director of GES NA Design & Automation, Amazon

Sumedha’s team develops a number of AI platforms.

GEMS AI MBR: Presentation Summary: GEMS Automations builds and deploy AI across the GEMS NA and INTL engineering portfolio, which represents over $40B in construction spend. Our goal is to help teams deliver engineering projects faster, at higher quality, and more efficiently, while helping Operations improve asset utilization (MHE uptime, associate productive time, DPMO, and maintenance cost). The new normal of building launches is not achievable unless we innovate to reduce rework and remove constraints across the engineering lifecycle. We do this through two complementary platforms, CORGI and Stella, governed by a single MBR mechanism that keeps AI investments disciplined, coordinated, differentiated, finance-validated, and non-duplicative.

What We Will Be Presenting

1. The two platforms in focus, CORGI (physical/perception AI) and Stella (agentic AI), and how together they cover the full facility lifecycle from construction through steady-state operations.
2. The measurable business impact of each, and the combined opportunity at scale.
3. Our framework for how we develop AI across the $40B+ NA and INTL portfolio, including the five-phase lifecycle (Ideation, PoC, Alpha, Beta, Rollout) and the governance that keeps the portfolio coordinated.

The Two Core Platforms

1. CORGI Unified Intelligence Platform
   1. An Amazon-developed Physical AI system that can see, understand, and act in live warehouse operations, integrating computer vision, PLC systems, micro-robotic arms, and robotic quadrupeds.
   2. Includes a Cognitive Twin, a digital replica of operations and equipment fused with AI reasoning, moving us from static dashboards to a closed-loop model that surfaces defects, finds root cause, and triggers the next action.
   3. Solves the siloed-systems problem (PLC, WMS, dashboards lack context) by unifying perception, reasoning, and prediction for real-time, proactive operations. Uses novel people-masking to focus only on packages, equipment, and process.
2. Stella Agentic AI

• A reusable agentic AI environment that supports engineering teams in the decisions required to design, launch, diagnose, and maintain MHE systems across Amazon operations.
• Not a single chatbot. It is a common product layer combining a text-to-agent builder, MHE-specific reasoning, custom engineering models, and a reusable connection layer to approved enterprise tools and data.
• Solves the fragmentation problem (documents, dashboards, vendor inputs, standards, project systems) that drove manual effort, SME dependency, and slow decision cycles.

Speaker BIO

Sumedha Katarya is Director of GES NA Design & Automation at Amazon, where she sits at the forefront of Amazon's efforts to harness artificial intelligence to accelerate execution, drive efficiency, and build the next generation of engineering capabilities — including the integration of AI-driven solutions that are reshaping how engineering work gets done.

With more than two decades of global leadership experience, Sumedha brings a rare blend of technical depth and strategic business thinking to the challenges of leading in an AI-first world. Her career spans engineering, product management, quality, and operational excellence across the energy and technology sectors — including senior leadership roles at TechnipFMC and nearly eight years at FMC Technologies, where she led product strategy, LEAN transformation, and new product introduction across global markets.

Sumedha holds an MBA from The University of Texas at Austin, a Master's degree in Civil and Environmental Engineering from the University of Massachusetts Amherst, and a Bachelor's degree in Chemical Engineering from the Indian Institute of Technology, Kharagpur.