Project Overview

ReiSearch is a real-estate-investment (REI) platform for research, deal analysis, and collaboration. I contributed as a core full-stack engineer across four codebases — from the original Python/Flask monolith through a full rewrite into React, Node serverless, and Go microservices.

What I built

• Token-metered billing: A reservation/settlement credits system — reserve on action, settle later — with recharge and shortage flows, implemented front to back.
• "Renovation AI" assistant: A threaded, streaming LLM chat feature (with its backing Go service) for renovation guidance.
• Real-time collaboration: Messaging and social-feed interactions over WebSockets, plus permissioned document/folder management and user connections & invites.
• Presentations & exports: Investor-ready presentation templates, PDF generation, and canvas-based editing.
• SEO link previews: OpenGraph meta-tag generation and media-URL normalization for shareable public links.

Technical implementation

• Frontend: React 19 + Vite + TypeScript, Redux Toolkit + RTK Query, Tailwind, Radix — with canvas editors (Fabric.js, Konva), data-viz, and guided onboarding.
• Serverless backend: Node/TypeScript on AWS Lambda (Serverless Framework), DynamoDB, S3, SES, SQS, and Cognito-authenticated JWT flows.
• Go microservices: A Go 1.23 + Gin service layer (16 domains, 23+ DynamoDB tables) with Prometheus metrics, multi-stage Docker builds, and structured logging.
• Origin monolith: The initial product ran on Python/Flask with Celery async jobs and Flask-SocketIO real-time chat.

Impact

The platform gives real-estate investors fast, accurate comps and market intelligence — and the billing, AI, and collaboration systems I built are the parts that keep it running in production.