The story behind OllamoMUI

OllamoMUI started as a simple idea: why pay $20/mo for ChatGPT when free models are just as good?

We built a lightweight API gateway that emulates Ollama, OpenAI, and Anthropic formats — routing every prompt to the best completely free LLM available. Then we added RAG, memory, a desktop app, a mobile app, and a full payment/licensing system. All open-source.

Rakibul Hasan — Rhasan@dev
Rakibul Hasan
@Rhasan@dev
Indie developer · Full-stack · AI/LLM
Star on GitHub ★
26
Free LLMs
181
Source Files
13
Test Suites
3
Platforms

Timeline

Day 1–2
Scaffolded the FastAPI gateway — Ollama-compatible /api/chat & /api/generate routing to free OpenRouter models. First prompt streamed end-to-end.
Day 3–4
Added RAG engine with pgvector + pg_trgm hybrid search, persistent memory with summarization, and rate limiting / audit logging middleware.
Day 5
Built the PySide6/QML desktop client with dual theme, chat playground, and RAG/memory browsers.
Day 6
Shipped the Next.js marketing site (23 pages), Lemon Squeezy payments, and license-key activation.
Day 7
Launched the React Native mobile app, Cloudflare Tunnel deploy, and rebranded everything to OllamoMUI — live in one week.

What's included

Backend (FastAPI)

  • ✓ 26 free LLMs via OpenRouter
  • ✓ RAG engine with hybrid search
  • ✓ Persistent memory with summarization
  • ✓ Multi-provider routing
  • ✓ Lemon Squeezy payments
  • ✓ License key generation & email
  • ✓ Rate limiting & audit logging

Desktop (PySide6 + QML)

  • ✓ Dual dark/light theme
  • ✓ Chat playground
  • ✓ RAG & memory browsers
  • ✓ Embedded terminal
  • ✓ Auto-updater
  • ✓ License activation screen

Mobile (React Native)

  • ✓ Chat, providers, RAG, memory
  • ✓ License activation
  • ✓ Usage analytics
  • ✓ Dark theme UI
  • ✓ Bottom navigation

Frontend (Next.js)

  • ✓ Marketing site with pricing
  • ✓ Public playground demo
  • ✓ RAG & memory demo pages
  • ✓ Payment result pages
  • ✓ SEO-optimized pages

🔒 Security Philosophy

When I built OllamoMUI, I started with a simple API wrapper. As the project grew, I realised that security couldn't be an afterthought. I implemented PBKDF2-HMAC-SHA256 password hashing with per-user salts, Role-Based Access Control with fine-grained permissions, and SSRF protection to block malicious requests. The desktop version bundles its own database, ensuring user data never leaves their machine. These security measures are not just features — they are part of a philosophy of data sovereignty and enterprise-grade protection. I am ready to bring this same level of architectural thinking to your remote team.

Built with

Python 3.14FastAPIPySide6QMLNext.jsReact NativeExpoPostgreSQLpgvectorDockerNginxVercelRenderNeonDBLemon SqueezyCloudflare

Hire Me

I'm Rakibul Hasan (Rhasan@dev), a full-stack developer based in Dhaka, Bangladesh (GMT+6), available for remote jobs, work-from-home, freelance, and local roles. I build full-stack applications, cross-platform desktop/mobile apps, AI/LLM integrations, and developer tools. If you have a project, an idea, or a problem to solve — let's talk.