🛡️ RegTech Compliance & Risk Screening Dashboard

An automated Governance, Risk, and Complance (GRC) and Open Source Intelligence (OSINT) platform designed to streamline corporate screening, anti-money laundering (AML), and know-your-customer (KYC) workflows. The application aggregates data from official state registries, open-source news feeds, and utilizes generative AI (Google Gemini 2.5) to deliver real-time risk assessments.

Key Features

• Multi-Row Registry Search: – Queries government databases via DaData API by Company Name or Tax ID (INN), handling ambiguous queries and displaying comprehensive legal entitles data (names, status, registration addresses).
• OSINT & Reputational Scraper: – Automatically scans media sources for adverse keywords and negative news regarding the target entity.
• Risk Scoring Engine: – Dynamically calculates a weighted risk score (0-100) based on sanctions listings, CBR blacklists, domain age verification, corporate status, and adverse media presence.
• AI-Driven Compliance Verdict: – Integrates the state-of-the-art Gemini 2.5 Flash model via the official 'google-genai' client to analyze negative context, offering structured summaries on regulatory, operational, and financial crime risks.
• Internationalization (i18n): – Features a full-fledged language switcher (RU/EN) within the interface, adjusting both UI text and LLM instructions dynamically.
• Local Auditing & Archive: – Automatically logs every screening session into an embedded SQLite database using standart SQL queries for historical compliance audits.

Tech Stack

• Language: Python 3.14+
• Frontend/UI: Streamlit Framework
• AI Orchestration: Google GenAI SDK (Gemini 2.5 Flash)
• Data Engineering: Pandas, SQLite3
• Integrations: DaData API (Rest Client)
• Environment: Fully containerized with VS Code DevConteiners, dependency management handled via 'uv' package installer.

Core Architecture Flow

1. Input: User submits an entity name or INN through the Streamlit UI.
2. Data Aggregation: Parallel data fetching from DaDataAPI + Local Sanctions/CBR Checkers.
3. OSINT Analysis: Extraction of adverse news context.
4. LLM Evaluation: Gemini 2.5 Flash processes the text array and outputs an expert risk verdict.
5. Persistence: Data is stored in SQLite; results are rendered dynamically on the web UI.

Prerequisites

• Python 3.9 - 3.14
• Docker Desktop (Optional, for DevContainers)

Quick Start

1. Clone the repository:

   git clone (https://github.com/abrenmarie/regtech-osint-finding.git)
   cd regtech-osint-finding

2. Set up a virtual environment and install dependencies:

   python -m venv venv
   source venv/bin/activate  # MacOS/Linux
   venv/Scripts/activate  # Windows

3. Install dependencies:

   pip install -r requirements.txt

4. Configure environment variables:

Create a .env file in the root directory and populate your keys

DADATA_API_KEY=your_dadata_key
DADATA_SECRET_KEY=your_dadata_secret
OPENSANCTIONS_API_KEY=your_opensanctions_key
GEMINI_API_KEY=your_gemini_api_key

5. Run the application:

   streamlit run app.py

Deployment in Streamlit Cloud:

To maintain functionally in production, specify the environment variables in the Streamlit Cloud dashboard under Advanced Settings -> Secrets:

DADATA_API_KEY=your_key
DADATA_SECRET_KEY=your_secret
GEMINI_API_KEY=your_gemini_key

License: This project is developed for educational and portfolio presentation purposes in the field of International Economics, Corporate Law, and RegTech Automation.