← All work
ARCHIVE · ENTRY 04 · CLASS: PUBLIC
Full-stack AI2025
LingualLink
Real-time multilingual chat
A real-time chat platform that translates across languages live, built on a production-shaped stack.
how it resolves
Message
Inference
Translation
WebSocket
Delivery
Problem
Real-time multilingual conversation means low-latency translation, persistent state, and presence — all at once, without the interface feeling laggy.
Built
A FastAPI + Next.js platform with WebSockets for live messaging, Redis for presence and caching, PostgreSQL and MongoDB for state, and optimized ML inference for in-line translation.
Role
Full-stack — backend services, real-time layer, data modelling, and the inference path.
Proof
- Live cross-language messaging with optimized inference.
- Production-shaped polyglot persistence (Postgres + MongoDB + Redis).
Stack
FastAPINext.jsWebSocketsRedisPostgreSQLMongoDB
Field note · classified