Skip to content

YakovBiton/MimicWork

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

64 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MimicWork

An AI-powered homework assistant that learns your writing style and generates answers that sound like you wrote them.

Live Demo

Note: 1. This demo uses paid AI APIs. To keep costs manageable, demo sessions are available by appointment (you can access the Vercel app to schedule a demo). 2. The backend may take 30-60 seconds to wake up on first request (free tier hosting).


What It Does

MimicWork analyzes your past writing samples to understand your unique style—vocabulary, sentence structure, tone, and complexity level. When you upload new homework questions, it generates answers that match your personal writing patterns.


Tech Stack

Layer Technology
Frontend React
Backend FastAPI (Python)
AI/ML OpenAI , text-embedding-3-small
Vector Database Qdrant (semantic search)
Deployment Vercel (frontend), Render (backend)

Technical Flow

  1. Document Processing — Uploaded documents are parsed and split into semantic chunks
  2. Embedding Generation — Each chunk is converted to a 1536-dimensional vector using OpenAI embeddings
  3. Vector Storage — Embeddings are stored in Qdrant for fast similarity search
  4. Style Extraction — When generating, the system retrieves the most relevant chunks from your writing samples
  5. Styled Generation — LLM generates answers using your writing samples as style references

Features

  • Style Learning — Analyzes vocabulary complexity, sentence patterns, and writing tone
  • Semantic Search — find relevant context
  • PDF Export — Download generated answers as formatted PDF documents
  • Multi-Document Support — Upload multiple reference documents for better style matching


Architecture Decisions

Why Qdrant
Qdrant provides a production-ready vector database with persistence, filtering, and a REST API—better suited for deployed applications than in-memory solutions.

Why OpenAI Embeddings over local models?
Render's free tier has 512MB RAM limit. OpenAI's embedding API eliminates the need for memory-intensive PyTorch models while providing high-quality embeddings.


Contact

KobbiLinkedIn | GitHub

Built as a portfolio project demonstrating full-stack development, AI integration, and production deployment.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors