Overview

The work built an end-to-end AI document intelligence pipeline for pharmaceutical SDFs, from OCR and layout extraction to RAG retrieval and a Gradio chat UI. Deliverables included extraction notebooks, OCR comparisons, RAG experiments, a multi-file Gradio interface, and a Colab demo with documentation. Evaluation results and retrieval/ocr metrics were recorded alongside architecture notes.

What I've accomplished

I produced a runnable, end-to-end document intelligence system for pharmaceutical SDFs that includes extraction notebooks, an OCR comparison report, multiple RAG configurations with evaluation, a Gradio RAG chatbot, and a Colab demo with measured performance.

Overview

The work explored building AI agents in n8n and assembling runnable workflows that produced images, merged marketplace datasets, and exported PDF reports. Deliverables included a Prompt Node with stored prompts for moodboard image generation, a Consumer Trend Discovery pipeline that produced a downloadable trend report PDF on area rugs, a Competitor Monitoring pipeline that generated a PDF

What I've accomplished

I produced working n8n agent workflows and artifacts: a Prompt Node with example prompts for moodboards, a Consumer Trend Discovery pipeline that exported a trend report PDF on area rugs, a Competitor Monitoring pipeline that generated a PDF comparing marketplace prices and offers, and an integrated Market Intelligence Dashboard with a slide deck.

Overview

As my graduation project, I designed and built Minterviewer — an AI interview prep and recruitment platform solving a real problem: unprepared candidates and slow, biased hiring. It delivers AI mock interviews, real-time performance analytics, and automated screening tools for companies. Built full-stack with React, Next.js, Node.js, and MongoDB, deployed on Vercel

What I've accomplished

Overview

Leading a 5-person team, I built MediCare — an award-winning clinic management platform solving real inefficiencies in healthcare admin. Features include real-time dashboards, secure patient records, and automated notifications to cut no-shows. Built with React, TypeScript, Firebase, and TailwindCSS. 🏆2ed Place, Gaza Sky Geeks.

Overview

Overview

The externship worked with CIYJ's multi-decade dataset of incident reports, surveys, demographics, staff records, and outcome measures. The work established a framework for appropriate analytical approaches, evaluated when and how correlational methods apply, and produced a clear scope of analysis for subsequent quantitative work.