Work — Vivek Bandi

Flux — AI Finance Tracker

Users needed a simple way to track spending and understand budgets through natural language interactions.

A finance tracker using Next.js (App Router), Drizzle ORM, and a natural language flow for logging and understanding expenses.

Real-time budget checks, overspending detection, and Supabase Auth with Row Level Security for user-level data safety.

Prioritized natural language UX clarity and simple budget tracking over complex aggregations and analytics features.

Next.js 16, Supabase (Auth + RLS), Drizzle ORM, Tambo SDK for natural language processing.

Building the interpreter required a reliable tokenization stage for source code analysis.

A Java scanner (lexer) handling keywords, identifiers, and operators for the Lox programming language.

Deterministic scanning rules and a clean token stream that can be consumed by a parser.

Focused on tokenization correctness and strict parsing over optimization and error recovery complexity.

Java.

Readers needed a quick way to search books and view details without a heavy interface.

A responsive React app integrated with the Open Library API for real-time search and retrieval.

Simple loading states, clear result rendering, and a lightweight interface designed for quick browsing.

Maintained simple search interface and lean feature set over advanced filtering, recommendations, or personalization.

React, Tailwind CSS, Open Library API.

The showroom needed a maintainable web platform with separate user and admin capabilities.

A full-stack electric vehicle showroom with browsing, offers management, and test-drive scheduling flows.

Node.js + Express backend, MongoDB persistence, and JWT-based Role-Based Access Control for admin access.

Implemented role-based access control for simplicity and maintainability over fine-grained permission management.

Node.js, Express, MongoDB, React, JWT authentication.