ML Engineer & Data Scientist
Applied Machine Learning
Generative AI
I build machine learning and generative AI systems: from a four-agent LLM pipeline with retrieval and 500+ automated tests, to clustering models with fairness audits built in, and computer vision pipelines that replace hours of manual work with minutes. Five years experience across consulting, higher education, and early-stage tech have given me range: I've been the researcher, the educator, the analyst, and now the person designing and building the ML system. My work stays grounded in the people the data reflects and supports.
01
Good Planning
Define
I work with stakeholders to turn an ambiguous problem into a clear question: what needs predicting, classifying, or measuring, and what decision the answer needs to support, before any model or pipeline gets built.
02
Problem Solving
Build
I design the models, pipelines, and evaluation frameworks the problem calls for: from clustering models and computer vision pipelines to LLM systems with retrieval and automated test coverage, plus the relational schemas and dashboards that get results in front of the people who need them.
03
Deliver
Inform
I translate model output and analysis into recommendations for strategy, risk, and where to intervene. My focus is helping teams decide what to prioritize and how to act on it, not just admire a metric.
I Help Organizations
From Data to Better Decisions
I’ve worked across applied machine learning, generative AI, and data science, in environments ranging from an early-stage startup’s founding team to international higher education to environmental-sector consulting. Across all of it, I use data and models to reduce ambiguity and improve decisions, for the organizations I work with and the people they serve.
Cross-functional collaboration
I regularly partner with leadership to align on metrics, validate assumptions, and deliver reports in a timely manner. This ensures insights are acted upon promptly.
Selected Work
Built a four-agent LLM system to turn free-text submissions into funder-ready impact reports for an international nonprofit, replacing an ineffective end-of-year form.
Built an image classifier for 16 tropical reef fish species, applying transfer learning, hierarchical classification, and data augmentation to underwater images from the iNaturalist API.
Built a two-page Power BI dashboard tracking student engagement, predictive readiness scoring, and application-to-hire funnel performance for an early-stage platform’s leadership.
Built a SQL (DuckDB) and Python pipeline that gives an operational system with no history a memory, enabling week-over-week and year-over-year institutional reporting.
Ranked affiliate study-abroad program enrollment by student major in Python to inform which partnerships justify further institutional investment.
Designed a normalized 9-table SQL schema and analytical queries to support scheduling, revenue, and capacity reporting for a tutoring business.
Applied unsupervised learning and a fairness audit to 4,424 student records, surfacing that financial burden was the strongest predictor of dropout risk.
Built a computer vision and OCR pipeline that automates question difficulty labeling at 98–99% accuracy, cutting manual review from 2 hours to under 5 minutes per test.
English (United States) 
English (UK)