All Projects

The Problem It Solves

Splitting costs in groups is painful. People forget who paid for dinner, spreadsheets get abandoned, and asking friends to pay you back feels uncomfortable. Existing apps like Splitwise are feature-bloated and slow. Let’s Split focuses on the core loop — add an expense, split it, settle up — with a native mobile experience that’s fast, offline-capable, and pleasant to use daily.

Key pain points addressed:

• No shared memory — groups lose track of who paid what over time
• Manual math — calculating each person’s share, especially for uneven splits
• Friction to log — most people don’t bother if it takes more than 10 seconds
• Context switching — manually tagging and categorizing expenses is tedious (the app uses a Bayesian classifier to auto-suggest categories)

Contracted to design and build the entire platform from scratch between January and June 2024. The goal was to replace a manual booking process with a fully self-serve online experience — tour discovery, availability checking, and secure payment in a single flow.

Backend & API: Built with Python and Flask, the backend handles tour catalogue management, booking lifecycle (enquiry → confirmation → payment), customer records, and a leads pipeline for the sales team. MongoDB stores tour packages, availability windows, and booking data in a flexible schema that the client can evolve without migrations.

Frontend: The customer-facing site uses HTMX for dynamic interactions — filtering tours, live availability checks, and booking forms — keeping the experience fast and server-driven without a heavy JS bundle. ImageKit handles responsive image delivery for tour photography.

Admin Dashboard: A separate Next.js dashboard gives the operations team full control over tour listings, pricing, availability, bookings, and leads — no engineering involvement needed for day-to-day content changes.

Integrations: Payment gateway for secure online transactions, Tawk.io live chat for customer support, and Google Tag Manager for analytics and conversion tracking.

Built to make sorting algorithms tangible. Instead of tracing through pseudocode or stepping through a debugger, you can watch the array transform in real time — each comparison, swap, and partition animated clearly so the logic clicks visually.

Algorithms implemented:

• Bubble Sort — repeatedly swaps adjacent out-of-order elements, surfacing the largest to the end each pass
• Merge Sort — recursively splits the array in half, sorts each side, then merges them back in order
• Quick Sort — selects a pivot, partitions elements around it, and sorts each partition recursively

The visualizer lets you see exactly which elements are being compared or moved at each step, making it easy to understand why one algorithm is faster than another depending on the input.