Mubeen Amjad
Home
About Me
Experience
Projects
Skills
People
Blog
Contact
Knowledge Graph
Ask my AI
Mubeen.AI
  1. Performance

Mubeen Amjad

Full-Stack & AI Engineer. Building SaaS products and AI features with Next.js, Node.js and Python.

GitHubLinkedInEmail

Portfolio

  • Projects
  • About
  • Experience
  • Skills

Connect

  • Blog
  • Contact
  • AI Agent
  • Status
  • Performance
  • Guestbook

© 2026 Mubeen Amjad. All rights reserved.

Visits are counted anonymously, without cookies (switch it off under site settings). Chat conversations are kept for up to 180 days, with technical data such as the IP address, to improve the assistant and prevent abuse.

Performance

How fast this site really is: a Lighthouse test of the live site every week, and the speed real visitors get. Nothing is filtered, including the heavy 3D pages.

Lighthouse (lab test)

The first weekly run hasn't happened yet.

Lighthouse loads each page in a fresh browser from a GitHub server; "mobile" simulates a mid-range phone on a slow connection, so it's stricter than most real visits. 90+ is good, 50–89 needs work.

Real visitors (Core Web Vitals)

75th percentile, last 28 days

Not enough visits yet. A page shows up after 5 measured visits.

Measured in visitors' browsers with the open-source web-vitals library and sent with the same anonymous visit counter (no cookies; nothing is sent if your browser asks not to be tracked). Good: LCP up to 2.5 s, INP up to 200 ms, CLS up to 0.1.

Why some pages are heavier

  • /brain draws a 3D knowledge graph with three.js, so it loads more JavaScript and works the graphics card. The other pages don't load it.
  • The talking avatar (a 7.5 MB 3D model, optional ~160 MB voice model) only downloads after you click it, so it doesn't slow down the pages themselves. Its speech runs in your browser, not on the server.
  • The site runs on one small shared server (about 1 GB of memory), which matters most for the first byte of each page.