Omni-Lab cover

AI/MLFeatured

Omni-Lab

Empirical AI Tutor with Three Specialized Agents

Source on GitHub (opens in new tab)Live demo (opens in new tab)

The problem

Static educational content (PDFs, slides, lecture videos) doesn't adapt to what a specific learner is confused about right now.

My approach

Three specialized agents — Teacher (live voice with barge-in), Scientist (writes code to verify claims), Visualizer (Veo-generated videos) — turn any PDF into an interactive, interruptible lesson.

What I built

Autonomous educational workspace powered by three AI agents: Teacher (Gemini 2.5 Flash Native Audio), Scientist (Gemini 3 Pro), Visualizer (Gemini 3 Flash). Upload any PDF → auto-generates slides, narrates via live voice, and lets you ask questions with real-time interruption. Generates videos via Veo.

How it works

Transforms static content into interactive, verifiable, visual lessons. Local VAD for barge-in voice interruptions. Scientist agent writes code to prove concepts rather than hallucinate. Flashcards, quizzes, infographics, and Veo-generated educational videos on demand.

What I'd do differently

The cleanest way to stop LLMs hallucinating in education is to let a Scientist agent write code that proves or disproves its own claims.

Good questions to ask

3 questions an interviewer could ask about Omni-Lab

  1. Why did you let a Scientist agent write code that proves or disproves its own claims?

    Based on: The cleanest way to stop LLMs hallucinating in education is to let a Scientist agent write code that proves or disproves its own claims.

  2. How did you design the Teacher agent to handle live voice with barge-in?

    Based on: Three specialized agents — Teacher (live voice with barge-in), Scientist (writes code to verify claims), Visualizer (Veo-generated videos) — turn any PDF into an interactive, interruptible lesson.

  3. Why did you choose to address static educational content that doesn't adapt to specific learners?

    Based on: Static educational content (PDFs, slides, lecture videos) doesn't adapt to what a specific learner is confused about right now.

AI-written from this site's content

AI-generated · may be wrongHow it worksPrivacy

At a glance

Technologies
7
Language
TypeScript
Updated
Feb 2026
  • React 19
  • TypeScript
  • Vite 6
  • Gemini 2.5 Flash Audio
  • Gemini 3 Pro
  • Veo 3.1
  • Recharts
  • #education
  • #multimodal
  • #voice-ai
  • #video-generation

Compare with another project