Findect

We spoke to many digital nomads and community companies in Bali to understand their needs and pain points. We stumbled upon Tyrone Williams who was a CEO and had the ability to match people who had problems and he would pair them up with people who could solve those problems. That was his whole business model. We also interviewed other companies who did the same thing, so we thought, "why don't we leverage this with technology and tap into this market?" Let's do it!

We built Findect, an iOS app that intelligently matches attendees based on their goals, professions, and interests. Our architecture combines a Node.js REST API for event and attendee management with a Python FastAPI service for AI-powered reasoning and vector search.

I served as the ios engineer, responsible for the entire iOS app. including app architecture, design, and implementation.

One challenge was ensuring that LLM-based reasoning matched real human intuition rather than superficial textual similarity. To solve this: • Engineered prompt templates that contextualized professional and personal alignment • Tuned retrieval hyperparameters for better diversity among top candidates • Added double filtering to prevent self-matches and redundant recommendations This project gave me deep experience in multi-service orchestration, semantic retrieval, and LLM-powered backend design.