@nlktech

A vector similarity search engine for e-commerce products. Instead of searching products by keyword, it represents every product as a numeric vector (price, rating, number of reviews, category) and uses FAISS — Facebook's library for fast vector search — to find the products whose vectors are closest to a query. You set filters (category, price, minimum rating, review count, how many results you want), and it returns the most similar products ranked by vector distance and a similarity percentage, instead of just filtering a table.

MamaHealth — a dual-track maternal and adolescent health platform serving pregnant women and adolescent girls separately, with strict row-level security isolating each track's data. Includes a RAG-based Q&A assistant grounded in WHO/MINSANTE documents via pgvector, so users get medically-sourced answers instead of generic chatbot output. A collaborator is building a companion vitals-monitoring device (temperature, heart rate, SpO2, BLE GATT blood pressure) to feed live readings into the platform.

I built a simple, clean Python tool that lets you edit images directly. It’s perfect for quickly adjusting brightness and contrast, so you can really bring out the details in your photos and make them pop. It’s a great way to handle image processing and makes editing visuals much faster and more interactive.