Hello world!

# This is the Microslac project

A clone version of Slack application with minimal functionalities, deployed on Kubernetes using Microservices architecture.

The backend HTTP service communicating with AWS Gateway WebSocket API
Receiving $connect, $default and $disconnect events from the Web UI, then write to key-value store Redis for other realtime workloads.
Consuming internal events, via RabbitMQ work queues, produced by other services and broadcast to interested client apps (identify by connection id).
Leveraging Redis keyspace notifications to implement realtime presence mechanism, in order to broadcast a particular user's status (online/offline) to other interested participants
Fully asynchronous implementation with FastAPI, FastStream, Redis Asyncio and SQLAlchemy Asyncio

Managing channels, members and messages generated from Web UI
Publishing various events (message.sent, member.joined, member.leave,...) to RabbitMQ
Developed using TDD method with testing libraries: Pytest, Factory Boy

FastAPI implementation as an intermediate layer before the inference with LLM models:
Serving model via HuggingFace TGI with the following models: Llama-3, Mistral-7B, Phi-2
Serving messages via LangServe and chat history with LangChain Memory

A read-only service for querying replicated data from PostgreSQL slave-database with SQLAlchemy
Data replication is implemented using Kafka Connect (specifically Debezium Source Connector and JDBC Sink Connector)