Game2Match
$ cat project_description.md
Game2Match is an AI-powered game curator that surfaces underrated “hidden gem” games based on a player’s vibe instead of store keywords. The user types a mood like “cozy horror, no jumpscares, short sessions” and Game2Match uses a FriendliAI LLM endpoint to parse that into structured preferences, score a curated dataset of indie titles, and return a mix-and-match set of recommendations that intentionally avoid the usual overhyped games. The frontend is a game-themed web UI (HTML/CSS/JS) that feels like a streaming service for niche games. The backend is a FastAPI service that talks to FriendliAI for language understanding and explanation, then logs each recommendation call to Comet/Opik for tracing and evaluation. The result is a small but complete vertical slice of an AI-driven game discovery product built specifically for players who are tired of scrolling endless store pages.
$ tech_stack --list
$ project_links --show
$ event_source --show
