SocialAI — Go Microservices Social Network With AI Images

What it does

People share photos and video, generate images from a prompt with DALL·E 3, and search posts by keyword, by author or by meaning.

Create a post or generate an image

How it is built

Front end: React + TypeScript.

Back end: Go microservices behind an Nginx gateway, with PostgreSQL as the source of truth.

Feeds: new posts reach followers' feeds through a transactional outbox, Kafka and Redis — so a broker outage delays a post but never loses or duplicates it. Retried uploads and messages carry idempotency keys, so a flaky connection never creates a second post.

Search: Elasticsearch, kept in step with the database by its own indexer.

How I know it works

The whole stack — PostgreSQL, Kafka, Redis, Elasticsearch, the gateway, the API services and the background workers — starts with one docker compose command. On every change, CI starts all of it and walks a real user journey through the gateway: sign up → follow → post → follower's feed → search → like notification → comment → direct message → delete. A separate job runs integration tests against real Elasticsearch, Kafka and Redis.

Stack: Go, React, TypeScript, PostgreSQL, Kafka, Redis, Elasticsearch, Docker

Code: https://github.com/Rolinmuuu/Social-AI-Frontend · https://github.com/Rolinmuuu/Social-AI-Backend

Personal project. Sample images in the screenshots are illustrations made for the demo.

0 likes · 4 views · Sep 26, 2026

A social network with AI image generation and semantic search, built as Go microservices that never lose or duplicate a post — with the whole stack tested end to end on every change.