🧩 Project Overview ​
Title ​
AI-Powered Taxation Assistant: Financial-Statement & Competitor-Based Tax Insights (RAG + LLM)
Motivation ​
Tax laws are intricate, and compliance often suffers due to:
- Complex tax classification of financial items
- Evolving Finance Acts and CBDT circulars
- Lack of automated benchmarking with peer companies
This project aims to bridge the gap between raw financial data and actionable tax intelligence through AI-powered automation.
Objectives ​
- Build a Retrieval-Augmented Generation (RAG) pipeline that integrates Finance Acts, CBDT notifications, and company filings.
- Classify assessee tax regimes (individual/company; presumptive or regular).
- Generate personalized tax compliance suggestions.
- Compute a risk-of-notice score based on anomalies or discrepancies.
- Validate outputs through expert evaluation and user studies.
Expected Deliverables ​
- Functional RAG+LLM prototype
- Finance Act summary report for recent FYs
- Evaluation metrics for accuracy and usefulness
- Recommendations for scaling and future integration with tax APIs