ProAudit AI
Personal Project
On-going
AI & Data Analytics
My Approach:
This project explores how data analytics and AI can automate audit workflows and improve financial risk detection.
Using structured financial datasets, I built a system that detects anomalies using statistical methods and generates AI-powered audit insights in real time.
To make insights actionable, I developed an interactive dashboard that visualizes financial trends, highlights risk patterns, and supports audit decision-making through clear, data-driven outputs.
Vision and Innovation
To build an intelligent audit system that reduces manual effort, identifies high-risk financial patterns, and delivers explainable insights through AI and visualization, helping auditors and analysts make faster, more accurate decisions.
Data & AI Process
Data Pre-Processing:
Cleaned and structured financial datasets; handled missing values and standardized numerical features.Feature Selection:
Focused on key financial indicators such as transaction values, frequency patterns, and variance metrics.Anomaly Detection:
Applied statistical methods (Z-score) to identify outliers and unusual financial behavior.AI Insights Generation:
Integrated OpenAI GPT models to generate contextual explanations and audit-ready summaries.
Tools used
Languages: Python (pandas, NumPy)
Frontend: Streamlit
Backend: Supabase (PostgreSQL, Auth, Row-Level Security)
AI: OpenAI API (GPT)
Visualization: Plotly / Streamlit charts
Data Handling: CSV, DataFrames
Version Control: GitHub
Key findings
Automated detection of anomalies across financial datasets using statistical thresholds
Real-time generation of AI-driven audit insights, reducing manual analysis effort
Identified patterns in financial data variability and risk indicators
Enabled multi-user collaboration with secure, role-based access to audit workspaces
