Back to catalogue
Project
Open TutorAI Contribution
Contributed to an Open TutorAI fork: a model allowlist for OpenAI-compatible providers, Markdown-free text for speech, a FastAPI adapter for NVIDIA image generation, and the Aflon Tutor AI rebrand.
How it’s builtHow it’s built
Open TutorAI Contribution · Contributions to an Open TutorAI fork rebranded as Aflon Tutor AI
In a fork of Open TutorAI I made 3 commits out of 383: a configurable model allowlist for OpenAI-compatible providers, a Markdown-stripping utility so the avatar's text-to-speech reads clean prose, a FastAPI adapter that lets the app generate images through NVIDIA's API, dev service scripts, and rebranding to Aflon Tutor AI across 42 files.
- commits authored by me
- 3 of 383
- distinct files changed across his commits
- 51
- lines added
- 400
- UI locales rebranded (ar-MA, en-US, fr-FR)
- 3
Technology and what it does here
- Python (FastAPI router)
- OPENAI_MODEL_ALLOWLIST filter applied in ai/providers/service.py and gateway/http/routers/providers.py
- FastAPI + httpx
- Devops/scripts/nvidia_image_adapter.py maps OpenAI image requests to NVIDIA flux.1-dev
- TypeScript
- Ui/src/lib/utils/tts.ts stripMarkdownForSpeech for avatar voice output
- Svelte
- AvatarChat/AvatarOverlay wired to the speech cleaner; branding across UI components
- Bash
- Dev-full.sh, run-service.sh, stop-dev.sh start and stop backend, adapter and frontend
- i18n JSON
- Rebranded strings in Arabic, English and French locale files
Implementation details
- The allowlist reads a comma-separated env var and filters provider model lists only when set, so providers that list models the account cannot run (e.g. NVIDIA free tier) stay usable.
- The image adapter converts NVIDIA's {artifacts:[{base64}]} response into the OpenAI {data:[{url}]} shape the app expects.
- npm scripts dev:full, dev:backend, dev:adapter, dev:frontend and logs:dev were added to ui/package.json (commit 081f9fa).
Technologies
- Open Source
- Model Configuration
- Text to Speech
