~7 years building payment, commerce and automation systems end to end: Java and Node.js services, React and Angular front ends. A Shopify & automation specialist for the Shopify, n8n, GoHighLevel and AI integrations around them, built AI-first, from spec to production.
vivek.sinha · intro
Tap for sound
3:04
Audio introduction from Vivek, about three minutes, with captions. Transcript:
I'm a full stack engineer who builds features end to end, from the back-end API to the screen the user sees, and ships them to production.
Seven years across FinTech, payments, retail and monitoring: Java & Spring Boot and Node.js microservices, React, Next.js and Angular front ends, Kafka, PostgreSQL and Redis underneath. Card and loan settlement at GSS, PayPal data-centre monitoring at Altimetrik, onboarding and offer generation at Accolite.
Lately I build headless Shopify commerce and the automation around it (n8n and GoHighLevel workflows, AI agents and chatbots), and ship it on Docker and Kubernetes. I work AI-first: Claude Code is my pair from the first spec to the production deploy, and I review every line it writes. The storefronts below are live: open them, switch devices, add to bag.
Experience
Seven years, shipped to production
GSSBengaluru Feb 2025 – Jan 2026
Senior Software Engineer
Credit-card, Fawry and Egypt payment modules plus early / partial loan settlement, delivered with zero major production defects.
Integrated the Bawaba settlement APIs with encrypted request / response payloads.
Migrated the Angular test suite from Karma to Jest; TDD to full functional coverage.
Productivity AI in the daily loop (Copilot, ChatGPT and Gemini for test scaffolding, refactors and API docs), plus reusable ThreadLocal backend components, so sprints closed faster with less repeated work.
Stack Java · Spring Boot · Node.js · React · Angular · PostgreSQL · Jest. Deployed to on-premise servers.
Java
Spring Boot
Node.js
React
Angular
PostgreSQL
Jest
On-premise
AltimetrikBengaluru May 2023 – Apr 2024
Senior Software Engineer · PayPal
Java + Spring Boot microservices for PayPal data-centre monitoring: sensor alerts over Kafka to consumer apps, hexagonal architecture, PostgreSQL outbox for reliable commits.
Plotly + Angular dashboards; APIs documented in Swagger.
Cut client-reported incidents by 20%.
Productivity Copilot and ChatGPT for boilerplate and test cases; hexagonal ports let adapters change without touching the core, and Java Streams parallelised the heavy data paths.
Stack Java · Spring Boot · Kafka · PostgreSQL · MySQL · React · Angular. Shipped as Docker images on Kubernetes (AWS).
Java
Spring Boot
Kafka
PostgreSQL
MySQL
React
Angular
Docker
Kubernetes
AWS
Accolite DigitalBengaluru Dec 2020 – Dec 2022
Senior Software Engineer
Onboarding and offer-generation backends in Node.js, Angular front ends with NgRx and RxJS.
JWT auth, pooled PostgreSQL access, a file-management service.
85% test coverage; system downtime down 25%.
Productivity TDD in Jasmine and Jest caught regressions before release, and connection pooling plus a shared file service removed repeated plumbing from every feature.
Stack Node.js · Angular · NgRx · RxJS · PostgreSQL. Shipped as Docker images on Kubernetes (AWS).
Node.js
Angular
NgRx
PostgreSQL
Docker
Kubernetes
AWS
Happiest Minds · Futorix Mar 2018 – Aug 2020
Software Engineer
Full stack delivery in Node.js and Angular: HR management, travel approval and vendor assignment systems.
Role-based JWT authorisation, NgRx state, TDD at 85% coverage.
Productivity Reusable Angular components, services and NgRx patterns shared across three products, so each new module started from working pieces.
Stack Node.js · Angular · NgRx · RxJS · PostgreSQL. Shipped as Docker images on Kubernetes (AWS).
Node.js
Angular
RxJS
PostgreSQL
Docker
Kubernetes
AWS
Across all of it: Redis caching that cut session latency by 70%, and CI/CD + TDD that shortened release cycles by 30%.
AWS Certified Developer – AssociateMicroservices: SAGA & Outbox10x AI ToolsB.Tech · Bharath University
Projects
Projects I’ve built
What each product does, what I built in it, and the stack it ran on.
Cookie & Consent Platform
Lets websites ask visitors for cookie consent and remember each visitor’s choice, so the site stays privacy-compliant.
Spring Boot APIs that store and return each visitor’s consent
React consent banner and settings screens, built on third-party consent libraries
Node.js middleware that applies the choice in the browser, covered by Jest tests
Front end
React
Back end
Java, Spring Boot, Node.js
Database
PostgreSQL
Deploy
Docker, Kubernetes, AWS
Healthcare Dashboard
A web dashboard where healthcare teams watch their key numbers update live as charts.
React screens with Redux, wired to the backend APIs for real-time data
Kept it fast and accessible (WCAG) on every screen
Unit and end-to-end tests on each feature
Front end
React, Redux
Back end
Node.js REST APIs
Database
PostgreSQL
Deploy
Docker, Kubernetes, AWS
Flexential Experience Platform (FXP)
The customer portal of Flexential, a US data-centre provider, where clients see their services and usage.
Reworked the front end in React 17 and Angular 8+ with NgRx state
Interactive usage charts with Plotly
Spring Boot and Spring Cloud microservices behind it, built test-first
Front end
React, Angular, NgRx
Back end
Java, Spring Boot, Spring Cloud
Database
PostgreSQL, MySQL
Deploy
Docker, Kubernetes, AWS
Buckman: Big Bang Platform
An IoT platform that reads sensors at industrial sites and raises an alarm when a value goes out of range.
APIs to register and set up devices
Collected telemetry from sensors and edge devices
KPI calculations and alarm rules, with unit tests
Front end
Angular
Back end
Node.js APIs
Database
PostgreSQL
Deploy
Docker, Kubernetes, AWS
Ascend: Assessment Platform (AACT)
An online assessment platform: authors build tests and content, learners take them.
Backend APIs for content and assessments
Unit tests on every API
Front-end screens to match the business rules
Front end
Angular
Back end
Node.js
Database
PostgreSQL
Deploy
Docker, Kubernetes, AWS
HR Management System
An internal HR system for managing employees and their HR requests in one place.
Node.js workflows that pull from several data sources
Angular modules with NgRx state
JWT login and safe database transactions (PgPromise)
Front end
Angular, NgRx
Back end
Node.js, JWT
Database
PostgreSQL
Deploy
Docker, Kubernetes, AWS
Travel Request & Vendor Assignment
Employees request business travel, managers approve it, and the system assigns a vendor and books the tickets.
Led both back end and front end
Request, approval and booking flow end to end
JWT login and transactional database writes
Front end
Angular
Back end
Node.js, JWT
Database
PostgreSQL
Deploy
Docker, Kubernetes, AWS
How I work
AI-first, from spec to production
AI is in my loop every working day, not as autocomplete, but carrying the work all the way to a verified release. Claude Code is my main pair; ChatGPT, Gemini and GitHub Copilot fill specific gaps. I stay the engineer: I set the architecture, review every diff, and the tests decide what ships.
A normal day
PlanClaude reads the codebase with me and we write the plan before any code: the root cause, the one right place to change, the files it touches.
BuildIt writes in small slices against a project rules file (design tokens, component library, security and tenant-isolation rules), so the output matches the house style instead of drifting.
VerifyIt runs the whole gate (lint, type-check, unit and Playwright tests, build), then checks the UI in a real browser at phone, tablet and desktop sizes before anything is called done.
ShipPull request, image build, GitOps deploy through ArgoCD, then a real request against production to confirm the change is live.
Building for a client, faster
Day-one prototypeRequirements become a spec, a data model and a clickable front end on the first day, so the client reacts to something real instead of a document.
Front end on railsA token system and shared components first; AI then generates pages against them, so a new screen is consistent from its first commit.
Integrations with testsShopify, payments, CRM webhooks and email wired with stubs and end-to-end tests that AI helps write. The storefronts here run 22 Playwright tests against a Shopify stub.
Content and mediaCopy drafts, imagery and voice: this page’s intro is my cloned voice (Cartesia), its captions timed with Whisper, the walkthrough narrated scene by scene.
Agents and toolsMCP servers connect the assistant to real tools (a browser, Shopify, voice and image generation), so it can do the work, not just describe it.
Proof on this pageThree Shopify storefronts with admin and in-place editor, a clinic-website farm, and this portfolio, all built with Claude Code, all deployed to Kubernetes.
My Claude setup
Connectors (MCP)Shopify for store data and GraphQL, Cartesia for voice and speech-to-text, Kling for images and video, Atlassian for Jira and Confluence, plus my own Shielva MCP server for bots, flows and knowledge bases.
BrowsersClaude’s built-in browser and Claude in Chrome: every UI change is opened, clicked and measured at phone, tablet and desktop sizes before it’s called done.
PluginsEngineering (code review, debugging, deploy checklist, system design) and Design (accessibility review, UX copy): the same checklist every time, not from memory.
Custom skillsMy house rules packaged as skills (ui-design, ui-refactor, shielva-connector), so a new page or connector starts from the agreed pattern.
Rules & memoryA project CLAUDE.md (architecture, design tokens, security, SOC 2 rules) and a running memory of past incidents and deploy steps, so mistakes are fixed once.
CLI & subagentsGitHub CLI for branches, PRs and merges; parallel subagents for wide code searches and independent reviews.
What I never hand over
Secrets and customer data stay out of prompts; credentials live in sealed config.
Security, auth and multi-tenant isolation are checked against written rules, not trusted.
AI output is verified like any other: a failing test, a second transcription model, a real browser.
Storefronts · headless Shopify
Three live shops on one platform
Built from scratch for three customer profiles: a cosmetics brand, a hair salon and a clothing label. Shopify runs products, checkout and payments; everything you see, plus the admin and its in-place editor, is my own code.
Shade finder that ranks by skin depth and undertone, a scroll-pinned shade rail, and a build-a-set bundle checked out as one validated cart.
Shade finder
Bundle builder
Reviews
Live site, fully interactive. Checkout uses Shopify's test gateway (card 1) and opens in a new tab.
How the Shopify side is built
A store and an app per brandEach brand is its own Shopify store with a custom app; my API trades the app’s client credentials for a short-lived Admin token, cached and sealed at rest.
Catalogue through Admin GraphQLProducts, variants, prices, stock and images seeded and edited through the Admin GraphQL API, with idempotent inventory and refund mutations.
Cart to hosted checkoutThe bag is a Storefront API cart; checkout hands off to Shopify’s hosted checkout and Shopify Payments.
Orders come back signedHMAC-verified webhooks sync every order, payment and refund into my Fastify + Postgres API, which drives order history, bookings and the admin.
Everything the shopper sees is mineNext.js front ends on each brand’s own domain, the admin and the in-place editor; a theme script sends “Continue shopping” back to the brand site.
Headless ShopifyStorefront API carts into Shopify checkout; Admin GraphQL for products, prices, stock, refunds; HMAC-verified webhooks keep orders in sync.
Next.js 16 · React 19App Router, TypeScript strict, server components, React Query, design tokens. Lighthouse 100 on every storefront.
Admin & in-place editorAdmins edit text, prices, photos, menus and sections directly on the live page; page builder with A/B tests.
Motion without librariesA small scroll engine (pinned stages, reveals) instead of framer-motion, and it respects reduced motion.
Shipped like productionPlaywright E2E, Docker images, Kubernetes via ArgoCD / GitOps, Cloudflare tunnel, one brand per domain.
Built AI-firstPlanned, written and tested with Claude Code as pair, every diff reviewed before it merged.
Freelance & contract
Four things, built to work together
Interface
Website Design & Development
Marketing sites, landing pages and web apps in Next.js and React. A real design-token system, accessibility to WCAG 2.1 AA, and Core Web Vitals that are still green six months after launch.
Operations
GoHighLevel Automation
Sub-account snapshots, workflows, pipelines and calendars set up so no lead sits waiting. Custom values, triggers and webhooks connecting GHL to the rest of your stack, not a pile of disconnected Zaps.
Intelligence
AI Automation
Enquiry triage, enrichment and routing backed by LLMs. Document and OCR pipelines, retrieval over your own knowledge base, and hand-off rules that know exactly when to stop guessing and fetch a human.
Conversation
Chatbots & Voice Agents
WhatsApp, web chat and inbound phone agents that qualify, answer and book directly into the calendar, with the full transcript written back to the contact so your team picks up mid-thread.
GoHighLevel
The whole client engine, in one platform
I set up GoHighLevel end to end (CRM, automations and funnels) and wire it to the rest of the stack with webhooks and custom values. Each capability below is a flow I build, node by node.
Core CRM & pipelines
Contact & lead tracking
Every contact and lead organised, and every deal followed from discovery to conversion.
Form · call · DM
Contact + source
Deal stage
Won
Visual Kanban pipelines
Custom sales pipelines that show exactly where each lead is and what moves it on.
New lead
Qualified
Proposal
Closed
SmartLists
Audiences filtered and segmented on tags and custom fields, ready for a campaign.
Tags + custom fields
Filter
Segment
Campaign
Marketing & workflow automation
Visual workflow builder
Multi-channel sequences built from triggers and actions, with waits and branches.
Trigger
Wait · if / else
Email · SMS · WhatsApp
Task to owner
Email & SMS marketing
Bulk campaigns, two-way SMS and MMS, with replies routed back to the team.
Segment
Bulk send
Two-way reply
Conversation
Unified conversation inbox
Email, text and social DMs in one thread per contact, assigned and answered in one place.
Email · SMS · DMs
One thread
Assign
Reply
Websites & funnels
Drag-and-drop builder
Landing pages, full websites and sales funnels launched without code.
Template
Landing page
Funnel steps
Live domain
Forms & surveys
Lead details captured straight into the CRM, each submission starting a workflow.
Form submit
Contact upsert
Workflow trigger
Booked
Memberships & courses
Digital products, communities and client portals with access granted on purchase.
Purchase
Access granted
Course · community
Client portal
Clinic websites
One codebase, a site per clinic
A live demo from the clinic-website farm: one deployment serves a branded site per practice from a registry row. Switch layouts and devices below; Northgate Family Health is fictional and its photos are AI-generated.
Family practices and small group clinics where patients pick a named clinician: a call-back form with this week's openings, named doctors, published prices.
Find a doctor
Call-back form
Price list
Live site, fully interactive. Use the demo bar inside to change colours or switch to Arabic.
How the clinic farm works
A clinic is a registry rowThe request’s host picks a row in MongoDB (name, theme, address, layout) and the same build renders that practice. A new site is an insert, not a deploy.
Trade and country do the restEight trades and sixteen country packs decide departments, treatments, prices, regulators and the right emergency number.
English and Arabic, same routesBoth languages come from one set of pages, right-to-left included, with nothing forked per locale.
Enquiries land in the CRMCall-back and booking forms are validated and rate-limited on the server, then forwarded to the clinic’s CRM webhook with a fallback.
Selected work
Systems I've shipped
Narrated walkthrough · 4 minAutomating Instagram and ThreadsA week of posts drafted, scheduled and queued across both networks from one console: per-brand voice, live phone preview, per-slot timing, and publishing volume tracked as it goes out.
Voice · CRM Healthcare
Clinic booking across three channels↗
One calendar, three front doors: an inbound phone agent, a WhatsApp thread and a web form. Each one checks live availability, holds the slot and upserts the contact, so double-booking isn't possible regardless of how the patient arrived.
Template authoring through to delivery: submission to the WABA, approval polling, per-tenant send credentials, and inbound replies routed to either the AI agent or a live inbox based on intent and confidence.
Agents drop a listing once. The pipeline syncs media, enriches the record, publishes to the site and the messaging channels, then captures every enquiry back against the original listing instead of a shared inbox.
A React admin surface where every query is tenant-scoped at the key, mutations apply optimistically and roll back cleanly on failure, and the whole thing runs off one token system so a new page never invents its own spacing.
A caller can interrupt mid-sentence and the agent stops talking, like a person would. Echo cancellation, voice-activity detection and streaming TTS keep the round trip short enough that the pause doesn't feel like a machine.
Hiring for a full stack role, remote or on-site? Let's talk.
Open to senior full stack and Shopify / commerce engineering roles: remote, hybrid or in the office (Bengaluru, Pune or Indore), full-time or contract. For freelance work, send a couple of lines about the business and I'll reply with how I'd build it, what it costs and how long it takes.
