Atlas
AI mentor that finds your tech track and plans how to learn it
A conversational career mentor. A chat works out which technical track suits you, a generated quiz measures your level, and Atlas plans a learning path from there.
Highlights
- Exploratory chat that locks in a track (ML, web, embedded…) from the conversation history
- Quizzes generated per track and graded into Beginner, Intermediate or Advanced
- Stateless FastAPI backend with strict Pydantic contracts; the Flutter client owns chat state
Flow
Chat → track selection → dynamic quiz → leveling → personalised recommendations. Each step is one endpoint, so the Flutter app can resume anywhere.
The first version ran Gemma 3 locally through Ollama. The backend was later moved to Gemini 2.5 Flash-Lite through the Google GenAI SDK.
Backend
FastAPI with Pydantic models for every request and response, written against an API contract shared with the mobile team.
The server holds no session state. The client sends the chat history with each call.