DoctorPy

DoctorPy is an AI assistant designed to support learning Python in a personalized way. Developed as a Proof of Concept during a workshop with Wild Code School learners and trainers, it served as an experimental basis before the launch of Space Coders Agency. Issues addressed: • Beginners need answers suited to their level • Recurring questions overload trainers • The lack of instant feedback slows progress • Generic resources do not adapt to the learner's context Teaching approach: • Answers adapted to the level (beginner, intermediate, advanced) • Contextualized, commented code examples • Cited sources to go further (official documentation, best practices) • Intuitive interface inspired by modern chatbots This PoC validated the value of educational AI with end users and shaped the architecture of Space Coders Agency.

Modern architecture based on independent microservices, each specialized in a specific function. Main services: • RAG service (port 8002) — Core of the system: generation of answers enriched by the knowledge base • Analytics (port 8003) — Tracking of usage and performance metrics • AI Quests (port 8004) — Gamification system for playful learning • Notifications (port 8006) — Personalized reminders and motivation Smart multi-level cache: • L1 (Memory) — Ultra-fast RAM cache (<1ms), 1,000 entries max • L2 (Redis) — Distributed cache for horizontal scalability • L3 (Disk) — Persistent SQLite cache (500MB max) • Adaptive strategies: LRU, LFU, TTL, SMART depending on the data type Semantic search engine: • Intent recognition (8 types: learn, how_to, debug, explain, compare, example, best_practice, general) • Smart autocompletion with relevance scoring • Extraction of Python entities (variables, functions, loops, classes) • Automatic complexity detection

The core of DoctorPy: structured answers adapted to the learner's level. Standardized answer format: • Explicit title of the concept covered • Accessible explanation with analogies (e.g. "a variable is like a labeled box") • Syntax-highlighted code blocks with comments • Important rules and pitfalls to avoid • Progressive practical examples • Sources available to go further Level adaptation: • Beginner — Simple explanations, no jargon, basic examples • Intermediate — Mid-level concepts, a few technical terms • Advanced — Complex concepts, patterns, optimizations Concepts covered: • Variables, data types, operators • Control structures (if, for, while) • Functions and parameters • Lists, dictionaries, tuples • Error handling (try/except) • Object-oriented programming

Modern user interface built with Streamlit, optimized for the learning experience. Tab navigation: • Home — System status, getting-started guide, sample questions • Ask a question — Chat interface with level configuration • Cache management — Statistics, optimization, maintenance • Search — Exploration of the knowledge base UX features: • Clickable frequently asked questions to get started quickly • Built-in level selector (beginner/intermediate/advanced) • Confidence and source indicators on each answer • Answer metadata (cache hit, processing time) • Professional dark theme Built-in monitoring: • Real-time service status • Cache statistics (hit rate, evictions) • One-click maintenance actions