Moto Gear Data Scrapper
Jan 2026 - Current (7 months)
MotoGearPick is a motorcycle gear comparison platform I built solo, like PCPartPicker but for helmets and riding gear. It's live at motogearpick.com. Users filter and compare across brands like AGV, Shark, Momodesign, and HJC by certifications, shell materials, Pinlock compatibility, closure type, and color family.
The backend is NestJS, TypeScript, PostgreSQL, and Redis on AWS (EC2, ElastiCache, CloudFront). Frontend is Next.js with SSR, chosen over a SPA because the product depends entirely on organic traffic. ORM is Prisma.
The hardest part was turning messy retailer pages into structured product data at scale. I built a dual-model AI pipeline: Gemini Flash extracts structured JSON from raw HTML, and Claude Haiku via AWS Bedrock handles