Skip to content
VS
All work
02 /

LectureAI

Lectures in, study material out.

Kind
personal project
Domain
INTELLIGENCE
LectureAI
Capture pending

Best asset for this section: a short silent screen recording (15–25s, muted, looping) of upload → transcript → generated notes. Prefer MP4 (H.264) plus a JPG poster frame.

/projects/lecture-ai/walkthrough.mp4

Overview

LectureAI takes an audio or video lecture, transcribes it, and generates learning material from the transcript — notes, quizzes and study aids.

The product surface is a subscription platform: accounts, cloud uploads, credits, billing through Stripe. Behind it, transcription and generation run asynchronously on GPU infrastructure, because a two-hour lecture cannot be processed inside a request.

Processing pipeline

  1. Audio / video
  2. Transcription
  3. AI processing
  4. Notes · Quiz · Study material

Technical challenge

The engineering constraint is that the expensive work is long-running and metered. Uploads have to survive the browser closing, jobs have to be tracked, credits have to be consumed exactly once, and transcripts are user data that needs to be stored carefully.

What it does

  • 01Authentication and account management
  • 02Cloud file uploads for audio and video
  • 03Subscription system via Stripe
  • 04Credit system metering AI processing
  • 05Secure transcript storage
  • 06Asynchronous AI processing on RunPod GPU infrastructure
  • 07Rate limiting, monitoring and audit logs

Technologies

Application

Next.js

TypeScript

React

Data

PostgreSQL

Prisma

AI & compute

OpenAI

RunPod GPU infrastructure

Commerce & operations

Stripe

Rate limiting

Monitoring

Audit logs

More

Transcript and generated notes
Screenshot pending

Still fallback used when the video is absent or on reduced-motion.

/projects/lecture-ai/transcript.png