work

since 2016

Experience

  1. 2022 – now

    Lead Data Scientist / Tech Lead

    CVS Health

    Medicaid Consumer Analytics & Behavior Change Team

    Lead a small DS team building Medicaid behavior-change interventions — $20M+ in active programs, internal tooling (launch-box, caliper), and the team's agentic-coding practice.

    • Lead a small team of data scientists running chronic-condition and medication-adherence interventions across the Medicaid population — programs the team owns generate $20M+ in annual medical cost savings.
    • Identified Fraud, Waste & Abuse in Rx as an untapped intervention area for the Medicaid population. Sized the opportunity at $93M–$310M annually using AMCP industry benchmarks (3–10% of spend) against our $3.1B internal spend, then narrowed with a team member to a $12.4M actionable spending pattern; first validating member campaign launching now.
    • Built launch-box, an internal bootstrapping tool for data scientists shipping GKE-deployed analytical apps: an interactive questionnaire generates a deployment-ready repo with infra configs, templates, and docs. Cuts time-to-market from weeks to hours by abstracting the platform layer DS shouldn’t need to learn.
    • Architected caliper, an evaluation framework for next-best-action campaigns — codifies impact measurement, automates monthly financial refreshes, and hosts campaign dashboards. Replaces per-campaign one-offs with a standardized workflow: better analytical rigor, less duplicated work, and DS time freed for new modeling.
    • Drive the team’s agentic-coding practice: designed a per-project flow where engineers iterate with Claude and trigger tailored Codex reviews on demand, with both agents grounded in committed CLAUDE.md / agents.md context. Established team-wide standards for code review, pre-commit/linting, environment management, and AI-assisted workflows.
  2. 2020 – 2022

    Senior Data Scientist

    CVS Health

    Medicaid Consumer Analytics & Behavior Change Team

    Shipped the team's first Rx adherence NBA and the foundational team infrastructure (Medipedia + shared Python library) still in use today.

    • Designed and launched the first Rx adherence next-best-action program for our Medicaid population — an XGBoost model identifying high-risk members for daily outreach. A randomized controlled trial measured a +2% lift in medication adherence (PDC), translating to $7M+ in annual medical cost savings.
    • Shipped additional NBAs across the chronic-condition portfolio — diabetes A1c testing, inhaler education, respiratory follow-up — each evaluated against condition-specific care-action and financial metrics. Diabetes A1c testing joined Rx adherence as one of the largest cost-saving programs still operating today.
    • Built Medipedia, a GitHub Pages knowledge base — onboarding, Medicaid data conventions, project documentation — alongside a shared Python library for common DS tasks. Both remain in active use; modeling internal infrastructure as a first-class deliverable inspired a teammate to start building libraries of their own.
  3. 2017 – 2020

    Data Scientist

    Domio

    Tech team

    First data hire at a short-term rental startup — built and owned the company's data, BI, and supporting backend stack.

    • Joined as the company’s first data hire; built foundational data infrastructure before DS work could land. Shipped 20+ ETL pipelines (BigQuery + Airflow + Pub/Sub on GCP) ingesting TBs from diverse sources, plus curated BI views and Chartio dashboards serving 100+ stakeholders. Selected Chartio after evaluating Tableau, Looker, and others against the company’s stage and budget.
    • Stretched into backend engineering to keep the stack running — APIs, DB migrations, deployment pipelines, and a JupyterHub deployment that standardized notebook-based runbooks across the org.
    • Built recursive web-scraping pipelines compiling a database of 1M+ U.S. short-term rental listings, feeding market-entry decisions (which cities to expand into) and pricing benchmarks against comparable properties.
    • Co-led hiring as the team grew — brought in a tech Lead (the team’s first manager), an additional DS, multiple engineers, and an SRE. Coached 30+ non-technical stakeholders on reading and acting on data, building a data-driven culture across the company.
  4. 2017

    Integrated Consumer Banking Solutions Team

    Bank-benchmarking analytics on Argus' multi-bank consortium data; introduced Python to a SQL/Excel-only team.

    • Built benchmarking analyses for commercial banking clients against Argus’ multi-bank consortium dataset — identified opportunities to grow deposit portfolios by millions.
    • Modeled and presented strategy recommendations to client executives — including a projected 18% lift in customer retention from P2P payments adoption, drawn from behavioral patterns across the consortium.
    • Introduced Python to a SQL + Excel–only team, streamlining batch workflows and onboarding 5+ colleagues to scripted analysis.
  5. 2016

    Fall Intern

    JPMorgan Chase & Co.

    Wholesale Credit Analytics and Solutions (WCAS)

    Fall intern with Wholesale Credit Analytics — SQL and Bash scripting on terabyte-scale credit data.

    • Scripted Bash automations to replace manual report uploads to remote systems.
    • Ran SQL queries over TB-scale credit data for risk-assessment reports; partnered with the quant research team to reconcile outputs and tighten technical documentation.

Education

  1. 2015 – 2016

    Master in Mathematics of Finance

    Columbia University

    • Example coursework: Time-Series Modelling, C++ Programming for Quantitative and Computational Finance, Numerical Methods in Finance, Multi-Asset Portfolio Management, Financial Risk Management.
    • Course project: Web-based Risk Calculation Engine.
  2. 2016

    Fellowship in Data Science

    The Data Incubator

    • Intensive data science bootcamp admitting fewer than 3% of thousands of applicants.
    • Weekly projects: identifying popular celebrities using web scraping and network analysis; analyzing restaurant inspection data with SQL; predicting Yelp score using scikit-learn; Wikipedia linguistic analysis via MapReduce.
    • Capstone project: analyzed NYC taxi and Uber trip data; built a web application to predict cab fare and trip duration given locations.
  3. 2011 – 2015

    Bachelor in Double major: Mathematics-Economics & Dance

    Wesleyan University

    • Honors thesis in Mathematics-Economics.
    • Dean’s List.
    • Completed the Dance major after starting to dance freshman year.