Hello, I'm

Shivang Bharadwaj

I'm a computer science professional with interests in Distributed Systems, Software Design, and Algorithms. I enjoy working with teams to build scalable, reliable software that makes a meaningful impact.

Seattle, WA

Experience

Amazon

Software Development Engineer II — Identity & Access Security

Sept 2024 – Present Seattle, WA
  • Architected the authentication and authorization framework enabling AI agents to securely access internal Amazon services, unblocking a $748M cost-saving initiative from inception through production launch.
  • Designed the consent experience, implemented role-based authorization, and drove cross-org alignment for Amazon's agentic identity infrastructure; authored vision documents, PRFAQs, and builder guidance shaping the multi-year roadmap.
  • Delivered a self-service trust management system for service-to-service authentication, enabling service owners to control inter-service communication through an approval workflow now powering multiple auth flows across Amazon.
  • Implemented critical security hardening: CSRF protection, IP-based access controls for satellite network clients, and automated credential rotation using AWS security primitives.

Amazon Web Services (AWS)

Software Development Engineer II — Data Center Automation

Apr 2024 – Aug 2024 Seattle, WA
  • Authored the technical design for a spare parts inventory tracking system integrating three AWS teams, defining the receiving and stowing experience across AWS's global data center fleet.
  • Implemented a blue/green deployment strategy with isolated testing stages for a critical asset management service, reducing blast radius during production deployments.
  • Investigated and resolved a 40-second page load regression in the asset management platform, delivering root cause analysis and remediation plan.

Amazon Web Services (AWS)

Software Development Engineer — Data Center Automation

Feb 2023 – Mar 2024 Seattle, WA
  • Redesigned the frontend architecture of the AWS data center operations platform, migrating to a shared component library that doubled test coverage across 800+ global data centers.
  • Delivered a workflow automation system that auto-populated 70% of change management inputs for data center technicians, deprecating redundant infrastructure and saving thousands in monthly hosting costs.
  • Discovered and fixed a critical authorization bypass that allowed unauthenticated users to execute data center operations, establishing access controls across all sites.

Amazon Web Services (AWS)

Software Development Engineer Intern

Jun 2022 – Aug 2022 Seattle, WA
  • Built a background job micro-service on AWS Lambda and DynamoDB handling 10,000+ work orders per minute across AWS data centers.
  • Migrated from a legacy EAM-based C# infrastructure to native AWS, eliminating deployment instability and reducing on-call tickets by 100%.

Microsoft

Software Engineer 2 — Word Web, Office 365

Mar 2021 – Aug 2021 Noida, IN
  • Founding team member of Microsoft's Noida engineering office; built on-call rotations, onboarding, and team structure that supported scaling from 20 to 5,000 engineers.
  • Led a full redesign of Word Web's UI to comply with Microsoft Fluent UI guidelines, with changes propagating automatically to PowerPoint and Excel.
  • Drove Word Web's accessibility compliance from 4% to 47% as measured by Microsoft's internal audit.

Microsoft

Software Engineer — Word Web, Office 365

Jul 2019 – Feb 2021 Noida, IN
  • Built the Word Web transcription feature end to end in React/TypeScript — audio upload, OneDrive integration, browser playback, speaker diarization UI, and document insertion — shipped to Word and PowerPoint for Microsoft 365.
  • Designed and built the Word Web Ruler as sole engineer using a canvas-based rendering approach to achieve sub-millisecond render times across resource-constrained browsers.
  • Contributed to the migration of Word Web's client-side codebase from Script-Sharp to TypeScript, significantly reducing build and test times.

Bharti Airtel Limited

Product Engineering Intern

Jan 2019 – Jun 2019 Gurugram, IN
  • Built a face liveness detection prototype for SIM card KYC; curated a ~15,000 image dataset by augmenting NUAA with Airtel customer data, resulting in a granted US patent (US11244150B2).
  • Improved Aadhaar OCR accuracy from ~75% to ~95% by building an image quality enhancement pipeline with card segmentation and field extraction.

Aspiring Minds

Software Development Intern

May 2018 – Jun 2018 Gurugram, IN
  • Built an automated proctoring system using OpenCV and Haar Cascades to detect facial landmarks across webcam snapshots, generating a confidence score to flag potential cheating during candidate assessments.

WorldBrain.io

Google Summer of Code Intern

May 2017 – Aug 2017 Remote
  • Selected for Google Summer of Code to build the testing and CI infrastructure for WorldBrain's open-source web annotation platform across a multi-repository codebase.
  • Built Jest testing infrastructure from scratch across 5 repositories, raising code coverage from 0% to 69%.

Education

Columbia University

MS in Computer Science

Aug 2021 – Dec 2022 · New York, NY

GPA: 4.00 / 4.00

Coursework: Deep Learning, Machine Learning, Artificial Intelligence, Computer Vision, Deep Learning Performance

Manipal Institute of Technology

B.Tech in Computer Science & Engineering

Aug 2015 – Jun 2019 · Manipal, IN

GPA: 9.01 / 10.00

Coursework: Data Structures & Algorithms, Operating Systems, Compiler Design, Parallel Programming, Computer Networks

Projects & Research

DeDrop — Attentive GAN for Raindrop Removal

Proposed a method for raindrop removal and generation from single images using Attentive GAN. Compared 3 methods with a top SSIM score of 91 on the test set.

PyTorchGANComputer Vision

COVID-19 Infection % Estimation

Built an ensemble model to estimate COVID-19 infection percentage from CT-Scans, X-Rays, and Ultrasounds. Achieved top 5 in MAE in the CodaLab competition.

PyTorchEnsemble LearningMedical Imaging

Fault Localization with Transformers

Researched AST representation for fault localization in Java code with unit tests on RoBERTa. Achieved pre-training perplexity of 0.125.

RoBERTaNLPProgram Analysis

Parallel Neural Networks

Implemented the HOGWILD algorithm with CPU-based parallelization for neural networks in PyTorch. Increased CPU usage from 12% to 60% with a 75% speedup in training time.

PyTorchDistributed TrainingParallelism

GoMLP & RNums

A multilayer perceptron library utilizing the Residue Number System (RNS), designed as a proof of concept for data path optimization in low-powered hardware.

GoNumber TheoryHardware ML

TOTP Implementation

Implemented Time-Based One-Time Password (RFC 6238) in Go with additional encryption extending the Constant Time Comparison module for enhanced security.

GoCryptographySecurity

Patents & Publications

  • US Patent US11244150B2 — "A Non-Intrusive Method of Face Liveness Detection using Specular Reflection and Local Binary Patterns." Filed March 2020 · Granted February 2022.
  • "Fast Base Extension in Residue Number System using Single Redundant Modulus." International Conference on Power Electronics, Control and Automation, Delhi, November 2019.
  • "An Exploration of FPGA based Multilayer Perceptron using Residue Number System for Space Applications." 14th IEEE International Conference on Signal Processing, Beijing, August 2018.

Skills

Languages

TypeScriptGoPython C#C/C++JavaScript SQLBash

ML & CV

PyTorchTensorFlowOpenCV Hugging FaceScikit-learnTF-GAN

Cloud & Infrastructure

AWS LambdaDynamoDBAWS IAM AzureGoogle Cloud

Frontend

ReactHTML5CSS3Canvas API

Databases

PostgreSQLMySQL DynamoDBMongoDBMS SQL Server

Tools

GitKubeflowDocker ApacheNginx

Get In Touch

Happy to chat about interesting problems, new opportunities, or anything else. Feel free to reach out.