← Work / DwellCasa
Client project 2026

DwellCasa

Hotel management platform for a boutique property in Lalitpur — listings, bookings, inventory, and a full admin panel.

DwellCasa — main interface

Overview

The problem

The client needed to run day-to-day hotel operations — listings, bookings, multi-location inventory — from one place, without paying for an off-the-shelf SaaS tool that would be over-engineered for a single boutique property.

What it is

DwellCasa is a complete hotel management platform built for a boutique property in Lalitpur. The client needed to manage listings, handle bookings, track inventory across multiple locations, and run day-to-day operations from a single admin panel — without paying for an off-the-shelf SaaS tool that would be over-engineered for their scale.

My role & approach

Solo client project — I designed and built the entire platform, including the multi-location inventory architecture, role-based access for front desk, manager, and admin staff, and Google OAuth onboarding.

Built with the repository pattern for a clean separation between data access and business logic, which made the multi-location inventory support significantly easier to reason about. Role-based access control ensures front desk staff, managers, and admins each see only what they need. Google SSO was included to simplify staff onboarding.

System flow

Client
Laravel
Repository Layer
MySQL

Laravel branches by role

Front Desk Manager Admin

"One admin panel for listings, bookings, and staff — instead of an off-the-shelf tool built for a scale they didn't need."

Stack

Laravel PHP MySQL Tailwind CSS Alpine.js Google OAuth

A closer look

Impact

3

staff roles with dedicated access — front desk, manager, admin

2026

shipped and live for a boutique hotel in Lalitpur

3

core admin modules — listings, bookings & staff

Multi-location inventory built into the core architecture

Next project

MotoVault

Semantic product recommendation engine using SBERT embeddings and cosine similarity — finds relevant bikes from natural language queries.