How chandraprakash-tripathi/ai_learning_companion works
Project type and purpose
• This is a full-stack, educational “ML playground” web application called AI Learning Companion, which uses Streamlit for the UI (frontend) and Flask for the ML-computation engine (backend).
• Its goal is to let users interactively explore gradient descent, activation functions, hyperparameters, and model comparisons via real-time visualizations.
File structure
• Top-level directories:
• backend/ – Flask app entry point (app.py), core ML logic (ml_models.py), helper utilities (utils.py), and plotting functions (visualizations.py).
• frontend/ – Streamlit UI files (0_Home.py), organized into components/ (reusable UI modules) and pages/ (each ML topic as a separate Streamlit page).
• tests/ – unit tests for the ML logic (test_ml_models.py).
• Config files: requirements.txt lists Python dependencies; no DB or external microservices are present.
README for architecture clues
• Confirms a two-tier architecture:
Streamlit UI ↔ REST API ↔ ML engine
• Tech stack: Flask for the API, scikit-learn/NumPy/Pandas for ML, Matplotlib/Altair for charts.
Main components and their relationships
a. Streamlit Frontend
• Entry point: frontend/0_Home.py
• UI Components: sliders, controls, math panels, sidebar (frontend/components/...).
• Pages for each feature (gradient descent, activation functions, etc.)
b. Flask Backend API
• Single Flask service in backend/app.py exposing endpoints (e.g., /gradient_descent, /activation).
• Receives JSON payloads containing user-adjusted parameters.
c. ML Engine & Visualization Modules
• ml_models.py: implements core algorithms (gradient descent, model training).
• utils.py: shared helpers (data generation, normalization).
• visualizations.py: wraps Matplotlib/Altair logic to produce plots or data arrays.
d. External Libraries
• scikit-learn, NumPy, Pandas for data and model computations.
• Matplotlib/Altair for chart generation.
Data Flow:
User → Streamlit UI (HTTP POST/GET) → Flask API → ML Engine (ml_models + utils) → Visualization module → Flask returns data/images → Streamlit renders plots
Architectural patterns & design principles
• Separation of concerns: UI logic is fully decoupled from computation logic.
• RESTful API: simple JSON-over-HTTP interface between frontend and backend.
• Modular code: backend ML logic split into cohesive modules; frontend UI split into components and pages.