Customer-facing web work for a D2C brand: a support chatbot, an offers and cohort-targeting tool, recommendations, and a faster, rebuilt frontend.
Country Delight is a daily-delivery D2C brand with a large consumer base. The customer-facing app needed to do more for that base: answer common questions without a support agent, put the right offer in front of the right people, and suggest what to reorder. It also needed to load fast on the everyday phones most customers actually use.
This is a live consumer product, so changes shipped against real traffic. Marketing needed to run targeted offers themselves without an engineer in the loop for every campaign, and the front end had grown heavy enough that load time was a real cost on mobile.
I built a rule-based chatbot that maps common support questions to guided answer flows and hands the edge cases to a person. For promotions I built the customer-criteria component: the team defines a cohort by attributes and behaviour, then applies an offer to exactly those users, with no engineering needed per campaign. I worked on the recommendation system that suggests what a customer might add next from their order history, and led a revamp of the customer-facing UI in Angular and Material-UI with a component-driven structure. Alongside that I made the build smaller and the site faster through code splitting, dynamic and lazy loading, image optimization, and performance profiling.
Customers self-serve more of their support through the chatbot, marketing runs targeted offers on its own, and the rebuilt front end loads faster on the devices most people actually use. Live at countrydelight.in.
The customer-criteria component behind promotions. The marketing team defines a cohort by attributes and behaviour, then applies an offer to exactly those users, so a campaign goes out without an engineer wiring it up each time.
A chatbot that maps common questions to guided answer flows and routes the edge cases on to a human, taking routine queries off the support team.
A recommendation system that suggests what a customer might add to their next order, based on what they've bought before.
Rebuilt the customer-facing interface in Angular and Material-UI with a component-driven structure, so the front end was consistent and quicker to extend.
Brought the bundle size down and the site up to speed with code splitting, dynamic and lazy loading, image optimization, and profiling out the slow paths. Load time matters most on the everyday phones the audience uses.
