• Prototype a large language model-driven AI chat toolo Use OpenAI and Pinecone APIs to build a custom retrieval augmented generation chat toolo Design and implement guardrails against hallucination, data insertion, and out of scope usageo Use Streamlit to build a customer-ready UI around the tool• Develop and productionalize an evaluation system for large language model chat exchangeso Guide client discussion sessions to identify areas of interesto Re-interpret targeted concepts as quantifiable metricso Engineer LLM prompts to score chat exchangeso Lead client team through human calibration exercises to validate evaluation results• Support client needs via data mining, predictive modeling, forecasting, and visualization

• Managed a team of five data scientists dedicated to predicting emerging clinical needs• Maintained and expanded existing analytic portfolioo Iterated on palliative care need model to make referrals available to an additional two million memberso Added to suite of post-acute care models with a model identifying appropriate episodic reimbursement payments for 100,000 skilled nursing facility stays each year• Improved performance of legacy predictive modelso Evaluated performance of legacy model predicting risk of hospital readmission; used R and H2O to cut scoring time by two thirds while maintaining equivalent performanceo Identified and delegated ten similar opportunities related to other utilization metrics

• Developed predictive model to identify ideal patient care journeys in the post-acute care space (R – sparklyR, H2O, and tidyverse)o Consulted with clinicians involved in each stage of patient care journey to understand process flow and opportunities for interventiono Developed a suite of models to generate care path recommendations at two key stages of care journeyo Designed and facilitated a weekly audit process to verify model performance with and solicit feedback from supervising clinicianso Incorporated machine learning model feature drivers in weekly audit process to build cross departmental understanding of machine learning modelso Developed and implemented small batch scoring to run every fifteen minutes with limited resources• Developed predictive model to identify members in need of palliative care services (Python – PySpark, sparkML, pandas, and scikit-learn)o Worked with clinicians to understand patient care journey and known wisdom around end of life careo Coordinated with vendor management organization to understand existing resources in the palliative space and identify gapso Joined forces with multiple other teams to implement regular model scoring and cohort development• Implemented daily scoring of vendor-developed model to predict risk of hospital readmissiono Liaised to third-party vendor to define target KPI and understand their model driverso Converted model scoring code from SAS to R (sparklyR and tidyverse)o Designed model performance evaluation mechanism and implemented routine evaluation procedures

• Assisted in development of clinical data ingestion and content extraction pipeline including natural language processingo Developed Python and Hive QL scripts to inventory and assess data available in existing text files and tableso Produced engaging and informative materials including Tableau dashboards and live demonstrations to educate business owners on the value of the available data and determine their needso Developed Python scripts to leverage natural language processing algorithms to meet key business needs from start to finish (e.g. extracting uncharted diagnoses from electronic medical record notes and prioritizing those likely related to care gaps)o Designed queryable tables to make key data accessible to low-level analysts elsewhere in the organization• Developed executive dashboard to provide at-a-glance insights to senior leadership and inform strategic business decisionso Lead senior leadership strategy meetings to identify areas of interesto Coordinated with business owners to inventory existing reporting efforts and develop sustainable data extraction planso Used Tableau to develop impactful and informative visualizations

• Identified, predicted, and tracked savings from cost management initiativeso Coordinated with clinical staff to identify potential medical cost savings and determine payment policy changeso Leveraged raw claims data to predict savingso Developed savings tracking mechanism and monitor program performance on an ongoing basiso Reported savings information to key stakeholders including actuarial team for reserve modeling• Determined value of existing contracts and impact of potential contract renegotiationso Combined medical claims, plan enrollment, and state revenue data to determine value of existing hospital network contractso Created dynamic reporting to show impact of various potential rate structure changes• Generated custom spend and volume reporting on an ad hoc basis

• Provided analytic support for design of customer outreach programo Developed a decision tree model to identify subgroups of likely responders to aid in targeting sales messageo Developed measurement mechanism for customer outreach programo Reported key findings to program stakeholders and at division-wide meeting• Measured impact of major retail network transformation on key customer experience metricso Coordinated with customer experience team to develop a data stream better suited to answering these questions

• Developed and maintained linear regression models to forecast client sales volume, order size, and other KPIs• Interpreted model coefficients to determine drivers of change and measure impact of each• Combined data mining techniques with traditional price elasticity and merchandizing efficiency modeling techniques to develop engaging presentation material for industry events • Prepared, presented, and interpreted results for clients with varying levels of statistical literacy

Independently edited and applied editorial conventions to evidence-based medical practice guidelines and prescription drug summaries

Developed and implemented a peer-to-peer advising system for first generation college students at Brown, and coordinated programming and resources for these students.

Provided in-class tutoring and support for struggling math students at Hope High School.

Taught SAT preparation classes for high school students; earned 100% satisfaction rating from initial class.