• TypeScript
• Gemini
• Firecrawl
• Serper API

Overview

An automated interview-question discovery engine.

• Searches, scrapes, and extracts company-specific interview questions from 16+ trusted sources, including Reddit, GeeksforGeeks, and LeetCode Discuss.
• Designed an AI-powered deduplication and ranking pipeline with rate limiting, exponential backoff, and source-weighted scoring to generate high-quality interview datasets.

Architecture

Overview

An automated data pipeline that discovers, scrapes, and curates company-specific interview questions into clean datasets.

Discovery & scraping

• Serper API performs targeted search across 16+ trusted sources (Reddit, GeeksforGeeks, LeetCode Discuss, and more).
• Firecrawl scrapes and extracts readable content from each result.

Extraction & curation

• Gemini extracts structured interview questions from noisy page content.
• An AI-powered deduplication and ranking pipeline removes near-duplicates and scores questions by source-weighted relevance.

Reliability

• Rate limiting and exponential backoff keep the crawler resilient against source throttling, producing high-quality, deduplicated interview datasets.