Open to Senior Backend Engineer roles

Hello there 👋

I'm Shipra Gupta.
I build
backend systems.

Backend-focused engineer, currently Software Engineer 2 at Microsoft, with 7+ years in software engineering across Azure and AWS: REST APIs and Redis caching for IoT payment devices, a framework behind 40+ microservices, AI tooling adopted across teams, and health features used by millions. This page is my cover letter, with receipts.

  • Microsoft
  • Paytm
  • Samsung
~/shipra: zsh
Java · C# REST APIs Distributed Systems Azure · AWS
7+years in software engineering
200→~80 msauth-ID lookup latency
~1M/daydatabase hits avoided
40+microservices on one framework

01 · cover_letter.md

Dear Hiring Team,

Instead of another cover letter, I built you this page.

I'm a backend-focused engineer. For 7+ years I have built software that has to hold up in production: REST APIs and caching for IoT payment devices on AWS at Paytm, health features on Samsung smartwatches used by millions, and today, at Microsoft, the frameworks and AI tooling that engineering teams rely on.

I like problems that compound. Auth-ID lookups with ~60% lower latency. A cache that saves a million database hits a day. A framework that removes ~95% of the manual work of standing up 40+ microservices. An AI agent that turns an hour of reporting into 90 seconds. I write the design doc, get teams in different time zones to agree on it, and see it through to production.

I'm now looking for a Senior Backend Engineer role where I can own services end to end. Scroll on for the receipts. One request on this page is still waiting for your response.

Warm regards,

Shipra Gupta

02 · Impact

Numbers, not adjectives.

Every project I've led ends with a measurable result. Here are some of them.

API latency · Paytm

80 ms

auth-ID fetch latency, down from 200 ms (~60% faster)

Database load · Paytm

~1,000,000

database hits avoided every day by adding a Redis cache layer

Microservice framework · Microsoft

40+ microservices (compute and non-compute) stood up through one framework I designed, with ~95% less manual work

AI agent · Microsoft

~95% less time to produce a sprint report, with a Claude-based agent working from live Azure DevOps data
Manual
30–60 min
With agent
~90 sec

Where my 7+ years went

7+years

    Three domains, one through-line: software that cannot afford to fail. Health data on your wrist, payment devices in shops, and the services Microsoft engineering teams run on.

    Platform migration

    100+

    build and release stages moved off a legacy system, following a design I authored

    Security by design

    ~40%

    lower deployment and security risk, plus ~10 hrs/week of manual review automated

    Search-backed Q&A · Microsoft

    ~50%

    less time spent finding answers in docs and wikis

    Before
    After

    APIs I built · Paytm

    • POSTdevice registration X.509
    • POSTtransaction notifications
    • GETtransaction summaries
    • GETdevice auth ID cached

    REST APIs for onboarding and authenticating IoT payment devices on AWS

    Samsung Health · Wear OS

    Millions of smartwatch users relied on the Stress and SpO2 trackers I developed during COVID-19

    03 · Selected work

    The receipts.

    Each project comes with a small interactive replay. Go ahead and press the buttons.

    Paytm · Backend on AWS

    A faster, safer backend for IoT payment devices.

    The problem. Large fleets of IoT payment devices needed secure registration, transaction notifications and fast authentication, without hammering the database on every request.

    • Built and optimized REST APIs for device onboarding on AWS, with secure registration using X.509 certificates.
    • Built APIs for transaction notifications and transaction-summary retrieval.
    • Added a Redis cache layer: about 1M fewer database hits a day, and auth-ID lookup latency down from 200 ms to ~80 ms.
    • Integrated AWS IoT Device Defender to detect abnormal behaviour and unauthorized access across the fleet.

    ~60% faster auth-ID lookups and a much lighter database.

    REST APIsRedisAWS IoTX.509Caching
    illustrative simulation · toggle the cache
    avg latency--
    cache hit rate--
    DB queries0

    Microsoft · AI agent

    A Claude-based agent that writes the sprint report.

    The problem. Putting together a sprint report from Azure DevOps took 30 to 60 minutes of manual work, and risks such as carry-overs and stuck PRs were easy to miss.

    • Built a Claude-based agent that works from live Azure DevOps data and produces the report end to end.
    • Automatically flags workflow risks: carry-overs, PBI bottlenecks and PR bottlenecks.
    • Ships 3 role-specific report modes from the same data, now used by multiple teams.

    ~95% less time: 30–60 min down to about 90 seconds.

    ClaudeAI AgentsLLM integrationAutomation
    sprint-report agent · illustrative replay at 20× speed
    Manual
    Agent
    $ press “Run” to start the replay

    Microsoft · Framework design

    One framework to stand up 40+ microservices.

    Designed and built a buildout framework that provisions the compute and non-compute resources behind 40+ microservices from declarative templates. Manual intervention dropped by ~95% and new services ship faster. I also wrote the runbook service owners use to onboard.

    illustrative · 0/42 services ready
    MicroservicesAzureARM TemplatesFramework design

    Microsoft · Search & retrieval

    ODSPDoc Buddy: ask the docs, get an answer.

    A Q&A chatbot backed by Azure Cognitive Search that answers questions from organization and project documents and wikis. Engineers spend ~50% less time looking for answers.

    Azure Cognitive SearchRetrievalChatbot

    Microsoft · System design

    Designing a 100+ stage platform migration.

    The problem. More than a hundred build and release stages still ran on outdated legacy infrastructure that needed modernizing, with stronger security and compliance.

    • Wrote the end-to-end system design, trading off scalability against security.
    • Worked with the China and US teams to agree on the approach and get the go-ahead for rollout.
    • Led the migration to standardized, reusable templates and retired the old infrastructure.
    • Built automated security and compliance checks into the new flow.

    ~40% lower deployment and security risk, and ~10 hrs/week of manual review gone.

    System DesignMigrationSecurityCross-geo collaboration
    design_doc.md · summary
    Goal

    Move 100+ stages off a legacy system onto standardized, reusable templates.

    Trade-off

    Shared templates for scale, with security and compliance checks enforced centrally on every run.

    Rollout
    1. Design doc reviewed
    2. China + US teams aligned
    3. Stages migrated
    4. Legacy retired
    legacylegacylegacy sharedtemplate 🛡security ✓compliance ⏵approval 🚀 every run passes the same checks

    Samsung · Wearables & secure storage

    Health features on millions of wrists.

    The moment. During COVID-19, people wanted to keep an eye on their blood oxygen and stress levels from their wrists.

    • Developed the Stress and SpO2 (blood-oxygen) trackers in Samsung Health on Wear OS, used by millions of Samsung smartwatch owners.
    • Worked on Samsung Pass, which securely stores app credentials and payment cards, and fixed API and UI issues to improve stability and performance.
    • As an intern, worked on adaptive brightness, which uses the camera to sense ambient light, improving user experience and battery life.

    Shipped at consumer scale, where every crash is someone's morning.

    JavaKotlinAndroidWear OS
    illustrative watch UI · not Samsung's actual design
    STRESS Low breathe in · breathe out BLOOD OXYGEN 98% ♥ 72 bpm

    04 · Journey

    My career, as a distributed trace.

    Every span returned 200 OK. The last one is waiting on you. Select a span to see its attributes and events.

    GET/careers/shipra-gupta
    trace_id 7f3a…c21 7 spans loading

    05 · How I work

    Four habits behind those numbers.

    01

    Design before code

    I write the design doc first. My platform migration design had teams in China and the US agreeing on it before anything moved.

    02

    Performance is a feature

    Find the hot path, then fix it. One Redis layer took auth-ID lookups from 200 ms to ~80 ms and took a million queries a day off the database.

    03

    Secure by default

    X.509 device auth at Paytm, secure credential storage at Samsung, and automated security checks at Microsoft. Security belongs in the design.

    04

    Automate the toil

    If I do something twice, I script it. If a whole team does it every week, I build a tool or an agent for it.

    06 · Stack

    The toolbox.

    Choose a domain to highlight the tools I use for it.

    07 · Beyond the 9-to-5

    I've been shipping since college.

    Side projects and wins from before anyone was paying me to build things.

    🏆
    All India Rank 10,558

    IIT-JEE Advanced 2015 (top 0.1%)

    🦊
    Published as a student

    Time Speaks went live on Mozilla Add-ons in July 2017

    🏸
    College badminton

    Represented LNMIIT at Desportivos, 2016 to 2018

    08 · Your move

    One request is still pending.

    POST/v1/interviews 202 Accepted · awaiting response
    • ✔7+ years as a software engineer at Microsoft, Paytm and Samsung
    • ✔API design, caching and performance tuning
    • ✔Microservices and cloud platforms on Azure and AWS
    • ✔System design docs and cross-geo alignment
    • ✔AI agents adopted across multiple teams
    • ✔Security by design: X.509 auth, secure storage, automated checks