Open to opportunities

Kevin Hennnessy

@kevinhennnessy

Message

Data Science intern focused on reliable multimodal ML and NLP for real-world decision-making.

United States

Message

What I'm looking for

I’m looking for an entry-level role where I can apply NLP and multimodal machine learning with strong evaluation and calibration—partnering with stakeholders to ship reliable data products, learn fast, and grow toward trustworthy AI and uncertainty-focused work.

I’m a Data Science intern building models that hold up in the real world—especially when data is messy, incomplete, and high-stakes. My work blends deep learning with careful evaluation so performance isn’t just strong, but trustworthy.

In my M.S. Data Science capstone (Multimodal Misdiagnosis Detection on MIMIC-IV), I led the time-series track and developed transformer-based sequence encoders for irregular clinical admissions. I engineered mask-aware preprocessing for sparsity patterns, built a production-grade evaluation harness, and improved calibration using temperature scaling, isotonic regression, and Platt scaling.

That project delivered AUROC 0.833, AUPRC 0.802, F1 0.746, and Brier 0.171—and it outperformed all individual encoders across primary metrics. When fusion collapsed, I diagnosed a loss weight misconfiguration in the CLUB mutual-information estimator (0.05 → 0.000001) and recovered calibration with flexible threshold optimization.

I also bring applied experience from internships: I built an unsupervised NLP pipeline to classify 120+ ambiguous IT finance help-desk categories, automated NLP preprocessing and interactive Power BI analytics for 1,000+ service requests, and supported enterprise data migrations using Tableau, AWS, Snowflake, and Oracle. I enjoy turning technical work into decisions—partnering with stakeholders, validating results, and shipping pipelines that others can use confidently.

Experience

Work history, roles, and key accomplishments

Data Science Intern

NetApp

May 2025 - Aug 2025 (3 months)

Built an unsupervised NLP pipeline to classify 120+ ambiguous finance help-desk categories into 5 actionable groups, evaluating embedding and topic modeling approaches to reach ~90% balanced accuracy. Partnered with finance stakeholders to validate results and support leadership-facing adoption.

Python NLP LDA Topic modeling Embeddings scikit learn ServiceNow

Operations Data Analyst Intern

NC Department of Information Technology

May 2024 - Aug 2024 (3 months)

Developed a Python-based NLP preprocessing pipeline for unstructured IT ticket data, automating text cleaning and tokenization workflows for trend visualization to support predictive resource allocation. Created interactive Power BI dashboards analyzing 1,000+ service requests for state IT leadership decision-making.

Python Power BI NLP SQL Tokenization Data Visualization

Digital Technology Intern

General Electric

Jun 2022 - Mar 2023 (9 months)

Migrated 100+ Tableau workbooks to a centralized enterprise data platform, improving data lake compliance by 30% and consolidating dashboards to improve usability and stakeholder access. Supported centralized reporting and access improvements through migration work.

Tableau SQL Data Migration

Enable Fellow

Fidelity Investments

Jun 2021 - Aug 2021 (2 months)

Validated migration of incident report data from Oracle to AWS using Snowflake and Tableau, confirming 100% accuracy across all migrated records. Ensured migrated datasets were complete and correct for reporting and downstream use.

AWS Snowflake Tableau Oracle Data Migration Data Validation

Education

Degrees, certifications, and relevant coursework

American University

Master of Science in Data Science, Data Science (Artificial Intelligence concentration)

2023 - 2025

Grade: 3.93/4.00

Activities and societies: Departmental Scholarship. Coursework: Advanced ML, Neural Networks & Deep Learning, Applied NLP, Statistical ML.

Earned an M.S. in Data Science with a concentration in Artificial Intelligence, completing coursework in advanced machine learning and applied NLP. Completed a multimodal misdiagnosis detection clinical ML capstone using irregular clinical time-series data.