Wildlife Trade Monitoring & Legal Classification Tool

Developed a modular tool to monitor and classify animal listings from online marketplaces (starting with OLX) for signs of illegal wildlife trade or mistreatment. The system automates the extraction, analysis, and flagging process using both text and image inputs.

Key capabilities:

Scraping: Extracts listing data (text + images) from platforms like OLX.

Custom LLM Classification: A locally finetuned language model evaluates listing text for legality under CITES, EU, and Romanian wildlife laws. Flags protected species, prohibited sales, and signs of mistreatment.

Species Identification & Taxonomy Matching: Combines image recognition and NLP to identify the species and match it to formal taxonomic databases.

Report Generation: Automatically generates legal documentation packages to support human review and potential complaints to authorities.

Built to support NGOs and animal protection groups in identifying and documenting illegal activity at scale—reducing manual review, improving precision, and creating a structured handoff for legal intervention.

Tools used: Python, BeautifulSoup, Selenium, custom finetuned LLM, image classification models, taxonomic databases Skills: NLP, ethical AI, web scraping, prompt engineering, legal logic modeling, automation