On VeriAgent, I built a document processing and inference workflow for invoice OCR, AI-assisted fiscal validation, arithmetic checks, and VeriFactu-style hash chaining. I integrated FastAPI services with PostgreSQL storage and a frontend workflow.
I also implemented a real-time data ingestion pipeline for Polymarket and SX Bet, with VWAP-based price calculation and event normalization. The project includes signal detection, Telegram notifications, and Docker Compose deployment.
For an Adult Income dataset project, I built a reproducible machine-learning workflow and compared Decision Tree and KNN models using scikit-learn pipelines. I used SHAP and permutation importance to examine model interpretability, and completed an internship at Professor Octopus AI Lab.
