Hi, this is Bhanu Kowshik: design engineer who designs and builds AI products, from problem to code. Currently building Verified Call.

Currently building Verified Call · see how I test it

Design engineerCurrently building Verified Call

I design and build AI products that scale, convert, and connect.

3 weeksto a first working Verified Call, built with AI agents
~2,000learners on EverTutor, where I was the sole product designer
  • Figma
  • Motion · GSAP
  • WebGL · Three.js
  • React
  • TypeScript
  • Next.js
  • Electron
  • Chrome extensions
  • SwiftUI
  • MediaPipe
  • Transformers.js
  • PyTorch
  • AWS Rekognition
  • Claude API
  • ElevenLabs voice
  • Agentic workflows

Selected work

Founder · designed, built and tested end to end, with AI agents

Verified Call

A check for video interviews: real camera, same person, no hidden AI.

  • 30/30caught, each of 4 hidden AI tools
  • 6 of 17attacks still open, published

5 hiring teams in talks · 3 weeks to a first working version

  • Electron
  • React
  • Chrome extension
  • MediaPipe
  • Amazon Rekognition
  • Deepgram
Read case study How I know it works

Founding designer · Figma to SwiftUI

Stressie

A stress check-in that pays for your evening.

  • 12% → 56%daily engagement in V2 pilots
  • $600Kfunding it helped secure
  • Figma
  • SwiftUI
  • Metal
Read case study

Sole product designer · EverTutor

EverTutor

A voice AI tutor: 1:1 tutoring for about $5 an hour, against ~$250 an hour for a human tutor.

  • ~2,000learners, 8+ schools
  • 104h → 4hto make a lesson, with Studio
  • Figma
  • Product design
  • Curriculum design
Live: the tutor Studio: how lessons get made

Sole product designer · EverTutor

EverTutor Analytics

Dense student data, turned into a real-time decision tool for educators.

  • Real-timestudent insights
  • Livein production
  • Figma
Request a walkthrough

Solo build · 48 hours

Dearly

Voice letters you can hear and keep.

  • 140+letters sent
  • 35five-star ratings
  • React
  • Framer Motion
  • Supabase
  • Vercel
Visit Dearly (opens in a new tab)

How I work with AI

I write the spec and make every design call. AI agents draft to it, and a separate agent checks the result before I accept it.

  • Claude Code and CursorVerified Call: a first working version in 3 weeks, with AI coding agents writing most of the code (Electron, React, Next.js) to my specs and test rules. This site is built the same way.
  • FigmaStressie: designed in Figma, built in SwiftUI, 13 of 13 frames matched in the build, with a lint that fails it on raw colours or sizes.
  • LovartImages and illustrations. I make images there rather than drawing them in code.
  • AI inside the productMediaPipe, Amazon Rekognition, PyTorch voice models trained on RunPod, the Claude API and ElevenLabs. The chat on this page runs on Claude, voice-cloned with ElevenLabs.
  • Evals before claimsPass/fail rules pinned before each test, exact 95% bounds, failures kept on record, and a build check that blocks product copy that overclaims. Published results ↗︎
Mail copied