In this project, I implemented a word similarity system using custom-trained word embeddings. The model takes a word vector (“chicken”) and returns the top 10 most contextually similar words based on cosine similarity.The results (like meat, fried, soup, pork) show that the system successfully groups words based on semantic meaning — not just spelling. This has real applications in search engines, recommendation systems, and NLP tasks where understanding meaning matters more than matching exact terms.

This project demonstrates my ability to:

• Train and query vector embeddings

• Build NLP tools for contextual similarity

• Apply cosine distance metrics effectively

• Translate raw data into intelligent search functionality

During the widespread CrowdStrike outage, I was on site at Mass General Brigham within hours of the systems going down to assist with emergency IT response efforts. I worked closely with technical teams and clinical staff to support recovery workflows, stabilize critical systems, and ensure communication channels remained functional. My role required fast thinking, adaptability, and a calm presence in a high-pressure environment, helping maintain continuity of care during a major infrastructure disruption.

I created a comprehensive cybersecurity risk assessment for a simulated health insurance company, focusing on protecting sensitive patient data and aligning with HIPAA compliance standards. The plan included detailed mitigation strategies, effort and cost estimates, and clear success metrics for security controls like MFA, data loss prevention, encryption, and vulnerability patching.

This project demonstrates my ability to:

• Assess industry-specific risks in regulated environments

• Develop practical, budget-conscious mitigation strategies

• Translate cybersecurity best practices into actionable plans

• Prioritize protections for high-value data such as PHI

​