Women & Money: A User-Centered Approach to Conducting a Personal Finance Product Discovery

Financial literacy remains a critical yet often overlooked skill among women. Recognizing this gap, I started the Women & Money Discussion Series — a multi-session program designed to help women gain confidence in budgeting, investing, economic trends, and long-term planning. From the outset, my guiding principle has been: build with the audience, not just for them.

🔍 Product Discovery Approach In part 1 of the study, I walk through how to collect user needs, validating feature hypotheses leveraging a cohort of ~40 women to shape a truly resonant experience, gather insights to understand product-market fit for a new personal finance product — especially one that leverages AI — and evaluate core technology (AI vs. SaaS) and distribution strategies (B2B vs. D2C).

❤️ Product Market-Fit Approach In part 2 of the study, I walk through enabling technologies analysis, distribution strategies (B2B vs D2C), defining a value proposition for your product (“moat”), and product roadmap.

Roles: Product manager / Brand Discovery / User Research
Technologies: AI, SaaS

A smart contract-powered lightweight journaling dApp using MetaMask Wallet and Base Sepolia that lets users post messages on-chain and react to them with ❤️ likes. Messages are displayed in a clean, emoji-enhanced card layout with a gradient background.

🔍 What is a Smart Contract?

Smartcontracts are programmable, self-executing programs, stored and run on a blockchain network, that automatically carries out the terms of an agreement when predefined conditions are met. Think of it like a digital vending machine: you put in the correct payment (condition met), and it automatically dispenses the product (action executed).

Roles: Product manager / Product designer / Smart Contract developer.
Technologies: Ethers.js, Hardhat, MetaMask, React, Vite, Basecan testnet for Smart Contract, & Vercel for UI

Content-Based Image Retrieval (CBIR) is a Python-based image search engine developed as part of my undergraduate coursework (B.Tech in Computer Science, 2016). Think of it as an early, simplified version of Google Image Search — without the reasoning or learning capabilities available in modern tools today (as of 2025).

🔍 What It Does

CBIR enables users to search for similar images by comparing visual features directly from the image itself — not using any metadata or tags. It uses color histograms as the core image descriptor to identify and rank similar images in a dataset using digital image processing concepts like Histograms of a digital image and Chi-squared distance algorithm to find similarity between images. Given a query image, the system returns a set of images from the database ranked by their similarity (from most to least similar) based purely on color proximity.

🧠 Key Features Searches images based on visual content (color histogram)
No metadata, tags, or text input required
Displays search results in descending order of similarity
Lightweight and easy to extend for other feature types (e.g., texture, shape)

Roles: Product designer / Back-end Developer
Technologies: Python, Anaconda, OpenCV