How gandhamvinaykumar/zepto-ai-ml-capstone-project works
Zepto AI/ML Capstone combines three independent workflows: book collection and relational storage, Titanic analysis and prediction, and policy-question answering through a local RAG service. In the data pipeline, one script owns scraping, cleaning, SQLite loading, and SQL examples. Analytics separates exploratory analysis from modeling and model reload. The support API routes policy questions through retrieval and answer generation, while general questions receive a direct response. The support implementation is sampled; some analytics modeling details and artifacts are documented but not present in the tree.
Main entry points: main.py Ask API and response · pipeline.py Scrape and clean books · 01_eda.py EDA and profiling · titanic.csv Titanic dataset
Components 16
Book pipeline
- Scrape and clean books
data_pipeline/pipeline.py - Books SQLite database
data_pipeline/pipeline.py - SQL query examples
data_pipeline/pipeline.py
Titanic analytics
- EDA and profiling
analytics/01_eda.py - Titanic dataset
analytics/titanic.csv - Cleaned Titanic data
analytics/clean_titanic.csv - Classification and regression
analytics/02_modeling.py - Model reload check
analytics/reload_pipeline_check.py
Policy assistant
- Ask API and response
support_assistant/main.py - Intent routing
support_assistant/main.py - General-question response
support_assistant/main.py - Policy retrieval and answer
support_assistant/main.py
Assistant data
- Policy documents
support_assistant/docs - Persistent embeddings
support_assistant/main.py
Other components
- Books website
- Questioner
Connections 16
- Books website → Scrape and clean books: serves pages
- Scrape and clean books → Books SQLite database: writes books
- Scrape and clean books → SQL query examples: runs SQL
- Titanic dataset → EDA and profiling: loads dataset
- EDA and profiling → Cleaned Titanic data: saves cleaned data
- Cleaned Titanic data → Classification and regression: loads data
- Classification and regression → Model reload check: saves pipeline
- Model reload check → Titanic dataset: reads sample
- Questioner → Ask API and response: submits query
- Ask API and response → Intent routing: invokes graph
- Intent routing → General-question response: routes general
- Intent routing → Policy retrieval and answer: routes policy
- Policy retrieval and answer → Persistent embeddings: queries embeddings
- Policy documents → Persistent embeddings: bootstraps documents
- General-question response → Ask API and response: returns response
- Policy retrieval and answer → Ask API and response: returns response