Keyword Finder is built on Django 5, FastAPI, Flask microservices, React 18, and GPT-4o. The platform supports 4 keyword generation modes — Seed, URL, Competitor, and Bulk. In Seed mode it builds prefix, suffix, and alpha combination lists, fires them to Google Autocomplete in parallel with rotating proxies, then runs a second deep expansion round before fetching volume, CPC, and competition data from the Google Ads API via OAuth2. SERP analysis jobs are distributed across a dedicated Flask scraping microservice fleet with round-robin load balancing and a four-tier fallback chain direct, Bright Data,
scraping.narf.ai, and CrawlBase — ensuring near-100% page retrieval success regardless of target site protection. Each competitor URL is processed through a full NLP pipeline using YAKE and NLTK, with domain authority pulled from Moz and Majestic via DomDetailer and cached locally to avoid redundant API calls.