How qiuethan/orbit works
Project type and purpose
• Orbit is a full-stack web application combining a React/Next.js frontend and a Python FastAPI backend.
• Its purpose is real-time facial recognition, audio capture/transcription, AI-driven topic extraction, web-based profile enrichment, and conversation history management.
File structure
• Top-level directories:
– backend/: Python services, subdivided into modules (facial_recognition, recording, search, llm, analysis_pipeline, cache, static, etc.) plus server.py and integration logic.
– frontend/: Next.js app with pages (app/api, dashboard, graph, vision), UI components (layout, main, profile, tasks, outreach), context, data adapters, hooks, and utils.
• Backend modules follow a pipeline/micro-kernel style: each concern (face recog, audio, LLM, web search) is encapsulated.
• Config files: backend/env.example, requirements.txt; frontend/package.json, next.config.mjs, eslint.config.mjs.
README for additional insights
• Core components and tech stack spelled out: React/Next.js, FastAPI, InsightFace/DeepFace, Groq Whisper, Cerebras LLM, FaceCheck.id, SERPAPI.
• Data flow: camera + microphone → backend pipelines → cache & logs → frontend real-time updates.
Summary
• Backend modules, their sequence in the pipeline, and interactions with external AI services and storage.
• Demonstrate real-time vs batch operations (WebSocket events vs end-of-session integration).
• Detail the separation between the React frontend and Python services while showing synchronous and asynchronous communication channels.