Data Science Intern
Octus
Jun 2025 - Jul 2025 (2 months)
• Automated the extraction of tabular data from high-volume legal fee court filings using Python-based packages to prompt engineer LLM powered systems. • Improved operational efficiency by reducing costs and processing time by roughly 65%, increased extraction accuracy by more than 10%, to meet rising demand from a 40,000 subscriber base. • Engineered statistical and business heuristic checks in collaboration with team members to ensure data validation and mitigate model hallucination.