E-commerce & AI Automation Developer · Siliguri, India · Open to remote UK roles
Abhishek Paul
I build AI automations that run inside real online stores: support chatbots, checkout systems and AI tools that handle live customer orders every day.
Before moving into development I spent 6+ years in performance marketing (Meta, Google and Amazon Ads), managing budgets up to ₹2.5 lakh a month. Alongside that I managed e-commerce stores and built websites on WooCommerce and Shopify. So I build automations with the business result in mind: fewer support tickets, fewer lost carts, more completed orders.
Claude & OpenAI APIsMCP serversGeminiWooCommerce / PHPReact NativeRazorpay
LLM tool-calling connected to order, shipping and product data.
Mobile apps
React Native apps for Android and iOS, from build to store listing.
E-commerce systems
Custom WooCommerce plugins, checkout and payment flows.
Growth thinking
Ads, funnels and conversion — I measure what an automation earns.
Selected work
Case studies
AI Shopping Chatbot
In production · v8.1
Problem
A D2C apparel store was answering the same "where is my order?" questions by hand. The answers sat across three systems: the store, the print-on-demand partner and the courier.
What I built
A WordPress chatbot plugin using Claude tool-calling
Live order lookup plus courier tracking via NimbusPost webhooks and API
Hourly sync with Qikink fulfilment data
Refund and return ticket filing, product recommendations and product Q&A
Customers get order status, tracking and product answers instantly, without staff time. While building it I worked out several undocumented courier and fulfilment API behaviours (webhook payloads, auth formats, order-key rules). I also built a companion reconciliation tool that diffs store orders against fulfilment exports.
Claude APIPHPWordPressWooCommerce RESTWebhookswp_cron
Store AI: WooCommerce MCP Connector
Hosted · mcp.cognicart.co
Problem
Store owners jump between dashboards to check sales, update orders, find low stock or create coupons. I wanted them to run the store by just asking an AI assistant.
What I built
A hosted Model Context Protocol (MCP) server that connects Claude to any WooCommerce store
Tools for sales reports, top products, customer insights, low stock, reviews, orders, products, coupons and bulk sales
Multi-store accounts with plan limits (Trial, Starter, Pro)
Tested end to end on a live store before launch
"How did sales do this week?"→Claude→MCP server→WooCommerce API→Report / action
Outcome
A working AI agent product. Owners can ask questions in plain English and take actions too, such as updating an order, editing a product or launching a sale. It shows I can build agent infrastructure, not just prompts.
MCPClaudeCloudflareWooCommerce RESTAuth & plans
Quick Checkout
In production
Problem
The store paid a monthly fee for a third-party checkout. It went down twice (a firewall block, then a 500 error from the provider's own API) and kept losing chargeback disputes.
What I built
A one-page popup checkout plugin on Razorpay, opened from "Buy Now" and from the cart
Partial cash on delivery: 20% paid online upfront, 80% on delivery
Coupon rules per payment type: coupons on full prepaid orders, none on partial COD
Buy Now / Cart→Popup: address + coupon→Razorpay (full / 20% COD)→Order created
Outcome
Removed a recurring fee and the outages that came with it. I tested my checkout against the paid provider on live orders, then switched fully to my own build. Partial COD cuts fake cash-on-delivery orders, because every buyer pays something upfront.
PHPJavaScriptRazorpay APIWooCommerce
AI Virtual Try-On
In production · web + app
Problem
Shoppers couldn't picture how a printed tee would look on them, which holds back purchases from new visitors.
What I built
An AI try-on: the shopper uploads a photo and sees themselves wearing the product
A backend endpoint that calls Gemini image generation, shared by the mobile app and the website
A WordPress plugin that adds the try-on to every product page
Origin locking, CORS handling and a 3-per-day limit per visitor to control AI costs
Product page→Shopper photo→Gemini image generation→Try-on result→Add to cart
Outcome
One AI feature serves both the website and the app from a single backend. Rate limiting and origin locking keep the cost of each try-on predictable and stop other sites from using the endpoint.
A D2C apparel brand wanted a native shopping app so repeat customers could browse, buy and get offers without going back to the website.
What I built
A React Native shopping app for Android and iOS, running on the store's WooCommerce APIs
Catalogue, cart, checkout and order history synced with the website
The AI try-on built into product pages
Push notifications, plus the Play Store listing assets (screenshots, feature graphic)
React Native app→WooCommerce REST API+AI try-on backend→Orders in one store
Outcome
Published on the Google Play Store. App and website share one product catalogue and one order system, so there's nothing extra to manage. I also tuned the cart API calls to make the app faster.
React NativeExpoWooCommerce RESTPush notificationsPlay Store
3D Esports Jersey Maker
Built · pre-launch
Problem
Esports teams and players want custom jerseys and team logos, but design tools are complex and there is almost no mobile app for it (I found one competitor on the Play Store).
What I built
A React Native app for Android and iOS where teams design a jersey on a live 3D model
Colour, pattern, name and number customisation, viewed from any angle
Team logo and mascot creation inside the app
Built with Claude Code as an AI pair programmer, and tested on real devices with Expo
Pick a template→3D jersey editor→Logo / mascot maker→Final team kit
Outcome
A working app on both platforms, built fast by pairing with AI coding tools. It is being prepared for launch on the Play Store and App Store.
React NativeExpo3D renderingClaude Code
Background
Experience
Aug 2026 – now
Freelance E-commerce & AI Automation DeveloperAI chatbots, MCP connectors, WooCommerce plugins, mobile apps and client websites
Apr 2025 – Jul 2026
Performance Marketing Specialist, Light Art StudioMeta and Google Ads
Nov 2024 – Mar 2025
Woobang IndiaPerformance marketing
Aug 2020 – Oct 2024
Gopaldhara TeaPerformance marketing and e-commerce