Kshitij Bhatnagar

I turn ideas into live products: the interface, the 3D, the AI and the infrastructure underneath. Six years of enterprise frontend by day, a studio of shipped side products by night.

projects built
24
live right now
13
years shipping
6
awards at work
5

01 / What I do

Four things I do well, and the work that proves each one.

01 / 04

Interfaces that feel effortless

Production React and TypeScript for enterprise clients since 2020: design systems, reusable component libraries, keyboard friendly accessible UI and pages that load fast.

  • Component library that cut dev time by 40%
  • 30 to 40% faster page loads
  • 85%+ unit test coverage held
  • React
  • TypeScript
  • Next.js
  • Web Components
  • SCSS
  • Tailwind
  • Jest
  • Accessibility

02 / 04

3D and data you can touch

Globes, particle fields and charts you drag. I use three.js, GSAP and d3 to turn dense numbers into something you understand by playing with it, like the particles on this page.

  • Live data globe for 260 cities and 193 countries
  • Tile free SVG flight globe
  • Draggable process capability curve
  • three.js
  • React Three Fiber
  • GLSL shaders
  • GSAP
  • d3
  • Chart.js
  • SVG
  • Canvas

03 / 04

AI that does real work

Agents, retrieval and speech pipelines with a human in the loop. I build with Claude, MCP, local models on Ollama, Whisper and Gemini, and I direct coding agents the way a lead directs a team.

  • MCP server that Claude and Cursor can call
  • Local RAG over 550+ personal memories
  • Nothing posts or spends without approval
  • Claude
  • MCP
  • RAG
  • Embeddings
  • Ollama
  • Whisper
  • Gemini
  • n8n

04 / 04

Shipped, secured and paid for

Products that run on real infrastructure, with sign in, payments, SEO and Cloudflare in front. I own the whole path from the database to the domain.

  • Google sign in and Razorpay payments in production
  • Cloudflare security on the product domains
  • Serverless Python, Cloud Run and Postgres
  • Node
  • Python
  • Firebase
  • Cloud Run
  • Postgres
  • Redis
  • Cloudflare
  • Vercel

02 / Work

24 projects, 13 of them live. Each one started with a real problem.

Every project

03 / Case studies

Problems I took apart to learn how they worked.

Each case shows how the work ran before, the turn I took, how small the setup is, and numbers that say where they came from.

  1. Timeline of the Post Aura case: four labelled tracks, Research, Website, Search and Product, with dated clips from 12 to 24 Oct 2026 and a cyan playhead on the last product clip.

    Video post production · 2026

    Post Aura

    Research, website, search, product

    Building a startup from scratch: the research, a website in five versions, its search layer and an AI footage tagger.

    Pre launch: site on test addresses, tagger tested on 10 real clips, schedule paused

    • 12 days from the starting wireframe on 12 Oct 2026 to the tagger's first production test on 24 Oct 2026, 23:15 IST. measured
    • 5 versions of the website over four calendar days, each frozen and kept live at its own address so the client could compare. measured
    • 10 clips from a real studio folder tagged in the first production test, 0 failed. A second run tagged none of them again. measuredTagged by the Gemini fallback model, not Claude. Failed counts errors, not wrong tags. Four camera files over 1,400 MB were skipped.
  2. Illustration on sample data: a drawing sheet with a lime scan bar over its title block, beside a transmittal where a new revision row is added and the old one is struck through in red.

    Architecture and engineering · 2026

    Drawing Ledger

    Workflow product on Microsoft 365

    The drawing fills in the transmittal. The lead still says yes.

    Built and tested, first live installation pending

    • 5 jobs to 1 per drawing issued: five hand jobs move to the engine, and the project lead keeps one review. reportedCounted from the product specification. Not yet observed on a live installation.
  3. A split flap board drawn for this page, at rest: STORIES FIRST on the top row and PRODUCTION NEXT on the second, ink letters on black tiles.

    Brand film production · 2026

    Flipp Tripp

    Research, design contest and build

    The studio had no showreel to lead with, so its first website leads with a split flap board of verified facts.

    Pre launch: review build waiting on the owner's confirmations; a requested new look has not started

    • 42 of 42 acceptance checks pass on the review build, reading the served pages with JavaScript off, 25 Oct 2026. measured
  4. The old and new Buskers home pages split down the middle. Left, the old page on 26 Sep 2026: its headline printed over the menu, above the About text. Right, the new page on 25 Oct 2026: the red LED dance floor under spotlights, the end of its headline and the Call now and WhatsApp buttons.

    Dance education · 2026

    Buskers

    Website takeover and rebuild

    I took over a dance school's 2018 website, kept it live until sign off, and broke none of its old page addresses.

    Live since 2 Oct 2026

    • 36 of 36 old page addresses land on the matching section of the new page, by permanent redirect. measuredChecked on 25 Oct 2026, each address with the trailing slash WordPress used.
  5. The header and hero of Shivantika Saxena's live site, with an arch drawing of a rose sun over sage hills and the heading “You don't need to have it all figured out. You just need a place to begin.” Below it, six steps joined by one rose line: read the page, choose a session, pick a time, answer her questions, pay her by UPI, join on Google Meet.

    Counselling practice · 2026

    Shivantika Saxena

    Therapy website and booking

    Her draft had a warm editorial look and no way to book; now clients book on their own and pay her by UPI.

    Live since 5 Aug 2026, booking checked 25 Oct 2026

    • 0 to 2 session types a client can book without messaging her first: a free 20 minute consult and a 50 minute session. measured

All case studies

04 / My personal automations

Small robots that run my day while I build.

Each one is a real workflow I built for myself, and the ones marked live are running today. Launch one to open its control room: drag the stages around, run the pipeline, make the human call, and knock an AI model offline to watch the fallback take over.

Live · every 5 min

A0120 stages · 27 data lines

LinkedIn Newsroom

Breaking AI news to an approved LinkedIn post, from my phone

57
sources polled
5
minute cycle
6
automatic drafts a day, max
0
posts without my tap

Live · every 3 min

A0215 stages · 20 data lines

First to Apply

New job postings screened, priced and on my phone within minutes

214
career pages
3
minute cycle
22,717
postings seen
7
screening questions

Live · every minute

A0314 stages · 22 data lines

Inbox Triage

Every email labelled, and my phone buzzes only for the few that need me

1
minute to sort
289
emails in the test set
93.1
% labelled right, simulated Jev
0.4
% wrong alerts, simulated Jev

Live · every minute

A0417 stages · 27 data lines

Claude Watchdog

Taps me when a Claude Code chat needs me or long work ends, and prices each automation's runs

1
spend ping per automation burst
4
Claude Code hooks
2
nudges per stuck item, max
0
pushes or deploys it triggers

Live · daily, on my Mac

A0528 stages · 37 data lines

Self-Improving Trading Lab

A paper money research lab that improves itself and is built to make fooling itself hard

Paper money only. The cloud side and the daily Mac runs are live. Most ideas fail, and nothing here promises returns.

20
counted ideas a week, max
7
kinds of strategy
3
months on paper before a change
1
finding a day, at least

05 / Journey

From circuits and GSM modules to enterprise frontends and AI products.

Experience

  1. HashedIn by Deloitte, February 2020 to present

    1. 2022 to Present

      Software Engineer II

      Technical Leadership & Innovation

      • Leading React.js development for enterprise applications
      • 40% improved data visualization accuracy
      • Built a reusable component library, 40% less dev time
      • Architected Microfrontend POC migration
      • Mentoring 5+ junior engineers
    2. 2020 to 2022

      Software Engineer I

      Foundation Building & Core Development

      • Multi factor authentication with OKTA
      • Secure PDF export and rich text editors with DraftJS
      • 85%+ unit test coverage maintained
      • 30 to 40% faster page loads
    3. Feb 2020 to Oct 2020

      Intern

      Rapid Skill Acquisition

      • Moved from HTML and CSS to ReactJS
      • React lifecycle, Redux and Context API
      • Built responsive PWAs
  2. May 2019 to Jun 2019

    Summer Intern · Ericsson India

    5G wireless research

  3. May 2018 to Jul 2018

    Summer Intern · Government of India

    Power plant efficiency studies

Recognition

Education

2016 to 2020

VIT University

B.Tech, Electronics and Communication, Minor in Computer Science. CGPA 8.8

2002 to 2016

The Shri Ram School

Higher Secondary, Science

Before software

  • 2019

    Mobile Phone System

    Calls, SMS and contacts on a Raspberry Pi with a GSM module.

    Raspberry Pi, GSM, Python

  • 2018

    Digital Lock System

    Password access control with attempt tracking and an emergency override.

    8051, EEPROM, keypad

  • 2017

    Universal Remote

    One controller that learns IR codes and switches mains appliances safely.

    IR, AC mains control

  • 2016

    Smart Home Security

    Motion and temperature monitoring with zone based alerts.

    PIR, thermistor

  • Under Secretary General, Internal Affairs, VITCMUN
  • Treasurer, SPIE Optics and Photonics Society
  • Technical Adviser, OWASP
  • Head of Marketing, TechnoVIT and Vibrance
  • Student Mentor, VITeach
  • Teaching Assistant, Circuit Theory
  • Organiser, Sociothon

06 / How I work

The projects show what I've built. This is how I build.

I direct AI the way a staff engineer directs a team. I frame the real problem, set the architecture, and make the trade-offs, then put the agent to work exploring the code, building, testing and shipping. The judgment stays mine, the speed comes from the machine. I've turned my process into reusable skills, so every project ships on the same baseline by default: strong SEO, Cloudflare security, and numbers I can stand behind. It's a way of working I've built on Anthropic's own agentic engineering practices, and it's why I can take an idea to a live, real-world product faster than a traditional build ever allowed.

products and experiments
24
live on the web
13
local AI apps
5
human in the loop
100%

07 / Contact

Have an idea worth building?

Products, frontend architecture, 3D on the web or putting AI to real work. Write to me, it goes straight to my inbox.