Marisol Delbrey

Work

Selected work.

  • Cellular Tower ML Model

    Fairness in AI Award

    Built a Python-based ML model to identify defective cellular towers across the U.S., translating business requirements into feature engineering specifications. Earned the Fairness in AI award by identifying systemic data bias and integrating underrepresented Puerto Rico data to improve both accuracy and equity of predictions.

    • Python
    • Pandas
    • Scikit-Learn
    Probability heat map of hurricane risk to United States cellular towers. Tower locations are plotted on a blue-to-red scale from zero to one, with the interior of the country in dark blue at near-zero probability and the Gulf coast, the Atlantic seaboard, Florida and Puerto Rico in yellow through red at the highest probabilities.

  • Collections Process Automation

    Production

    Built an end-to-end AR automation pipeline pulling from the firm's data warehouse, plus a dashboard and automated reminder flows. Along the way, identified and resolved a significant invoice tracking gap caused by a silent data sync failure. It reconciles to the source system to the cent, runs 44 automated tests, and refuses to publish if the totals disagree.

    • Python
    • Node
    • Power Automate
    • Supabase
    Overview of the AR Hub dashboard, showing aged receivables broken out by department and work type. Summary tiles across the top carry total open AR, past due and not yet due balances, with an aging mix donut and ranked bar charts by department, work type and leadership position below.

  • Client Wiki

    In Progress · Piloting

    Designed a centralized client-data dashboard to support the firm's advisory work, currently piloting with top-tier clients ahead of a year-end rollout.

    • Supabase

  • Firm-Wide Intake & Ops Rollout

    In Progress

    Replaced a multi-handoff intake process with a form that writes directly into Karbon via API, creating a single automated path with a full audit trail. Currently extending the integration to write to the CRM simultaneously.

    • Karbon API
    • Dynamics 365
    • Power Automate

  • Automation & Reporting Flows

    Production

    Built and maintain a set of recurring Power Automate flows spanning CRM reporting and the intake/API pipeline.

    • Power Automate
    • Dynamics 365
    • Karbon

  • Balanced Scorecard / KPI Program

    Production · Ongoing

    Own monthly KPI and scorecard reporting across two departments; led a 12-month historical rollforward and built the reporting infrastructure from scratch.

  • Accenture Innovation Challenge

    Finalist · Top 0.05%

    Advanced to the final round of Accenture's Innovation Challenge, developing a data-backed business strategy for the National Park Foundation. Defined KPIs and measurement frameworks through cost-benefit analysis and Gen Z user segmentation, producing impact projections that shaped the final recommendation presented to senior Accenture stakeholders.

    • Business Strategy
    • Cost-Benefit Analysis
    • Presentation Skills

Earlier experience

Healthcare Administration Intern Kaiser Permanente Surfaced inefficiencies in member data workflows using Excel and SQL, and coordinated cross-functional project milestones. May–Aug 2025 Internship
ML/AI Fellow Break Through Tech (UCLA/Cornell) Applied machine learning with Pandas, Scikit-Learn and NumPy, translating model output into stakeholder-facing reports. May 2023–Apr 2024 Fellowship
Student Research Assistant Chapman Fowler School of Engineering Modeled California water budget scenarios from 1998-2019 and visualized future supply and demand in Tableau. Jun–Dec 2023 Research

Want the longer version of any of these?

Ask me about it