JourneyAI
LIVEProduction-grade AI sales assistant platform serving 10+ enterprise clients with 98% uptime, developed over 8 months across EverythingToGain and Skaled (ongoing). Built sophisticated RAG pipeline with multi-collection vector search, real-time WebSocket streaming, session snapshots/forking system, and comprehensive subscription management. Contributed 25% of codebase including semantic search engine, background worker system, and Stripe integration.
Highlights
Real-Time AI Chat
12+ specialized AI assistants with session-aware WebSocket streaming, event batching, and token-by-token function call visualization.
RAG Pipeline & Semantic Search
Multi-collection vector search across Artifacts, Messages, and Files using OpenAI embeddings and Qdrant. Score thresholding, multi-tenant filters, and dynamic context injection.
Session Snapshots & Forking
Conversation branching with save states, fork from any message, and restore snapshots. Full history with vector embeddings for cross-session context.
Multi-Tenant Architecture
FastAPI backend with async patterns, organization-level data isolation, ARQ job queue (Redis), Loguru logging, and Docker containerization.
Dynamic Component Engine
JSON-to-React rendering system with type-safe registry supporting 70+ components. Backend-driven UI updates without frontend deployments.