dishant.dev

./about

About

I'm a Distributed Systems Engineer with experience keeping fleet-scale infrastructure running at Microsoft Azure. At Azure Core Compute, I work on bare-metal provisioning and disaster recovery for machines that run Cosmos, Azure Storage, XBOX, Exchange Online & others. Before that, seller and technician-side services at Amazon Home Services, and the Simulink engine at MathWorks. I'm drawn to the foundational layer of distributed systems: fault tolerance, recovery, the stuff that has to hold up at 1 MILLION+ machines and 3am pages.

Bengaluru, India dishant.sheth@hotmail.com download résumé

Stack

Languages
JavaC++C#SQL
Systems
Bare-metal provisioningFault-tolerant distributed systemsDisaster recoveryEvent-driven architectureWindows internals
Cloud & Ops
AzureAWSControl-plane engineeringFleet-scale operationsDC operations

Experience

Software Engineer - II (L62) — Microsoft — Azure Core Compute

Jun 2022 — present

Bare-metal provisioning and disaster recovery for fleets running Cosmos, Azure Storage, XBOX & others. Recipient of the Azure Core Culture Award for Customer Obsession.

  • Increased Azure Cosmos fleet availability by ~40%, leading an architecture overhaul of the JBOD provisioning stack to support multi-disk fault tolerance (up to 80% disk loss per machine), replacing a legacy single-disk failure model
  • Engineered a zero-touch BitLocker/TPM disaster recovery system from scratch for Azure Storage's data plane machines, cutting recovery time from days to ~10 minutes across 100K+ machine clusters
  • Overhauled bare-metal recovery architecture for Exchange Online, replacing indiscriminate full-disk repave with targeted faulty-disk recovery, cutting machine downtime from 1 hour to 20 minutes across 20K+ daily recoveries
  • Awarded the Azure Core Culture Award for Customer Obsession for a provisioning package that detected and remediated BitLocker protection gaps, restoring Australia clusters after a critical Office 365 outage
  • Resolved a fleet-wide watchdog instability causing transient failures for ~5 years by tracing a legacy WMI inefficiency via custom instrumentation and kernel dump monitoring, cutting WMI overhead by ~90%
C++,C#,Bare-metal provisioning,Fault-tolerant systems,Disks & storage systems

Software Engineer - I — Amazon — Home Services

Jun 2021 — May 2022

Built event-driven platforms and quote management tooling for seller and technician-side workflows in Amazon's home services business.

  • Unified fragmented notification services into a single event-driven platform, cutting onboarding time for new notifications by 50% across seller and technician workflows
  • Built Quote Lifecycle Management end-to-end, enabling Amazon's multi-million ARR consultative services program. Received an Amazon Accolade for the delivery
Java,AWS - EC2, DynamoDB, Lambda, SQS

Software Engineer - I — MathWorks — Simulink

Jul 2020 — Jun 2021

Built authoring tools for the Simulink Icon Editor, used by engineers to design and customize block masks.

  • Engineered LaTeX equation support with live preview in Simulink Icon Editor, enabling engineers to author block masks using mathematical notation. Shipped to production and available in Simulink today
  • Drove ~70% improvement in Icon Editor discoverability by embedding it into multiple block authoring workflows across Simulink
C++,JavaScript,Simulink,MATLAB

Education

M.Tech (Integrated) Software Engineering

VIT Vellore

Aug 2015 — May 2020