Projects — Ayush Mandowara's Portfolio

A curated selection of open-source projects and personal experiments spanning GenAI agents, full-stack apps, machine learning, NLP, automation tools, and developer infrastructure. Each project represents real-world problem-solving with production-grade code. Several repos (like Facebook-Auto-Liker and WhatsApp Web) have gained community traction with 45+ and 13+ stars respectively. Explore the repositories to see implementation details, commit history, and community contributions.

Mercari Japan AI Sho…

LLM-powered AI agent for Mercari Japan — scrapes products, stores in PostgreSQL, auto-tags with SEO metadata, and serves via Streamlit UI with hybrid English/Japanese query support. Uses OpenAI/OpenRouter with automatic fallback.

1
0
6

Python

Sarah's AI Recruiter

AI-powered recruiting tool built with Vite, React, TypeScript, and shadcn-ui. A modern full-stack application for intelligent talent acquisition.

0
0
8

TypeScript

YouTube Companion

Upload assistant and file viewer for YouTube channel @thevibecoder69. Built with Python and Streamlit to manage and organize channel content.

0
0
3

Python

YouTube Downloader

Streamlit app to download YouTube videos in 4K (or max resolution). Supports audio-only downloads using yt-dlp and ffmpeg with real-time progress tracking.

0
0
5

Python

Scan2Contact

SmartScan Contacts — a Next.js web app that extracts contact details from images using OCR.space API and converts them into structured digital contacts with Firebase backend.

1
2
7

TypeScript

Video Gesture Recogn…

Deep learning project: CNN + RNN (MobileNet + GRU) to recognize 5 hand gestures for smart TV control (thumbs up/down, swipe, pause). Achieved 92% validation accuracy with a ~40MB model optimized for embedded hardware.

1
0
24

Jupyter Notebook