As an AI / Data Science Intern at Developers Hub, I built end-to-end Python machine learning workflows, from raw data through trained and validated models. I also developed pipelines to ingest structured data from JSON, REST APIs, and HTML sources.
On InvestSafe, I built a FastAPI backend that analyzes investment documents against a 21-rule fraud framework and returns a risk score, risk level, and triggered red flags. Its detection pipeline combines deterministic regex rules with a LangGraph system of five specialized sub-agents.
I engineered document understanding with PyMuPDF and selective OpenAI vision OCR for scanned pages. I also designed asynchronous job processing, pluggable local and AWS S3 storage, and pytest-based evaluation, and deployed the project to Render.
My other projects include a Scikit-learn customer churn classifier tuned for imbalanced data and an AI-powered CV screening application. The screening app uses FastAPI, PostgreSQL (Supabase), and Celery to score and rank CVs against job descriptions.
