AI / Full-Stack Engineer — Community Intelligence & Analytics Dashboard for Game Studio
Orçamento: $20.0
FIXED /
⭐ 0.00 (0)
Uzbekistan
python, machine-learning, data-visualization
AI / Full-Stack Engineer: Discord Analytics Pipeline & Dashboard
The Problem
Our Discord community generates thousands of messages daily. Crucial signals — like viral cheaters or silent bugs — are lost in the noise and reach our team too late due to manual tracking.
The Goal
Build an automated pipeline that ingests raw chat data, classifies it via LLM, and presents actionable insights (JSON-structured and summarized) on a clean web dashboard.
Core Features
Targeted Ingestion: Incrementally pull data exclusively from explicitly authorized Discord channels.
LLM Classification: Output strict JSON to tag sentiment, bugs, cheating, and performance.
Cheater & Bug Tracking: Auto-detect, rank by confidence, and group scattered complaints into priority issues.
AI Summarization: Generate readable daily reports per category using a secondary LLM pass.
Live Dashboard: Interactive UI for QA/Community teams with trend charts and adjustable timeframes (7 to 180 days).
No-Code Tuning: Adjust custom game slang and context via editable prompt files without code changes.
Data Integrity: Idempotent, append-only design to prevent double-counting upon re-runs.
Tech Stack & Deliverables
Backend: Python, LLM API (Gemini/Claude/OpenAI)
Database: PostgreSQL
Frontend: Vue.js / Nuxt.js, Tailwind CSS
Deployment: Linux/Ubuntu (Containerized)
Deliverables: End-to-end working pipeline, live dashboard, and documentation for adding games/tuning prompts.
Roadmap (Bonus): Reddit/forum integration, dynamic risk weighting, and human-in-the-loop feedback to improve accuracy.
To Apply
Please share examples of similar LLM pipelines or data dashboards you have built. Briefly explain your architecture approach for guaranteeing structured, reliable JSON outputs from LLMs.
Abrir na Upwork