Sanket Gautam.
47.6062° N · 122.3321° W
SEATTLE, WASHINGTON
AI AGENT RELIABILITY · AMAZON · UW ALLEN SCHOOL

SanketGautam

Agentic AI and distributed systems. 8+ years building in production.

Software engineer at Amazon and a 2026 MS graduate of the University of Washington's Allen School. I spent eight years learning how large distributed systems fail, across Alexa AI, Buy with Prime, and Selling Partner Services. I now put that to work on the reliability and security of autonomous agents: measuring how long-running multi-agent systems fail, building primitives that contain those failures, and keeping sensitive work on-device. Read the write-up →

MS CS UW Allen School, 3.9 GPA 8 yrs distributed systems at Amazon Agentic AI shipped to production Poster presented, Data Conclave, Seattle Tech Week 2026
FEATURED WORK

Built and running

Four things I point people to first. Three of them link to the real thing.

ALSO BUILT

Lumora AI UW · GEEKWIRE

AI CRM platform for wealth managers, built in UW's software entrepreneurship course. Pitched to 16 VCs at Pioneer Square Labs and featured in GeekWire.

Read on GeekWire ↗

Project Panacea PEOPLE'S CHOICE

Multi-agent AI system for partner lifecycle management on AWS Bedrock. People's Choice Award at an Amazon hackathon, 2025.

Embedded Apps SDK AMAZON

A developer platform for running partner apps inside Seller Central, with an Intuit QuickBooks integration over PKCE OAuth.

EXPERIENCE

Eight years of systems

All at Amazon, from Hyderabad to Seattle, with earlier stops in research and health tech.

JUL 2020 - PRESENTSEATTLE · HYDERABAD

Software Development Engineer II

Amazon · Selling Partner Services, Buy with Prime, Alexa AI

  • Conceptualized and delivered the organization's first production agentic AI system, from PRFAQ through security certification to launch. Led a junior engineer to build an agentic tool that keeps CI/CD pipelines fresh automatically, and presented it org-wide as "Agentic AI at Amazon" (40 live, 160+ views).
  • Executed a partner-qualification service migration across 3 stacks with zero incidents, recognized with a Leadership Principles Award (Q4 2023) for the project and overall performance.
  • Led the Seller Central Embedded Apps launch across 4 teams in 2024: architected the Embedded SDK for data exchange, auth, and telemetry, and shipped the Intuit QuickBooks integration with PKCE OAuth.
  • Led the Operational Readiness Review for the Buy with Prime Returns 1.0 launch: 3 teams, 12+ engineers, 15 systems, 120+ technical requirements.
  • Cut new-customer onboarding for the Alexa data-labeling platform from 60 days to under 10 with a headless transformation layer.
  • Prevented dozens of potential high-severity incidents through proactive investigation, and eliminated recurring monthly downtime in a legacy ML task-routing system.
JUL 2018 - JUL 2020HYDERABAD

Software Development Engineer

Amazon · Global Selling, Selling Partner Network

  • Built Elasticsearch-backed search for an international listings platform, serving 1M+ product offers from thousands of global sellers.
  • Shipped an SSO-like migration moving sellers to Seller Central, integrating AuthN and AuthZ with marketplace authorization.
  • Developed configurable two-level distributed caching with exponential backoff and jitter for currency conversion.
2016 - 2017HYDERABAD · JODHPUR

Earlier roles

Amazon · DRDO · elth.ai

  • Amazon SDE intern: prototyped a seller-provider communication tool and automated localization testing for Global Selling.
  • DRDO technical trainee: FFT optimizations for CT image reconstruction.
  • elth.ai chatbot developer: symptom assessment and appointment booking for a medical chatbot serving 100+ hospitals.
ABOUT

The longer view

Sanket Gautam at dusk, with the skyline and Lake Washington behind him
Seattle, June 2026 · the evening I finished at the UW Allen School

I grew up in India, studied computer science at NIT Bhopal, and joined Amazon in 2018. Eight years later I am still drawn to the same problem: systems that stay up, and the quiet engineering that keeps them that way.

In 2023 I started a master's at the University of Washington's Paul G. Allen School while working full time, finishing in June 2026 with a 3.9 GPA. That stretch pushed my work toward agentic AI: the first production agentic system at my organization, and Ensemble, an independent multi-agent system I have kept in continuous production so I learn the operational failure modes firsthand rather than from a paper.

I was the first in my family to go into this field, so I spend real time mentoring: first-generation CS students at UW through GEN1, hackathon teams at DubHacks and LA Hacks, and young women across Asia through a UNESCO Technovation lab. Seattle is home; the skyline above is the view I picked it for.

EDUCATION
  • MS, Computer Science · University of Washington, Paul G. Allen School · 3.9 GPA · 2023 to 2026
  • BTech, Computer Science · Maulana Azad NIT Bhopal · 2014 to 2018
RECOGNITION
COMMUNITY
  • Judge, LA Hacks 2026 · Mentor and judge, DubHacks '25
  • Mentor, GEN1 CSE Community, UW, 2024 to present
  • Security certifier, Amazon Information Security, 2020 to present
  • Mentor, UNESCO Technovation Idea Innovation Lab