• Providing independent consulting for machine learning and data science projects leveraging healthcare data

• Technical lead of ML projects which included ideation with stakeholders, building ETLs to generate feature engineered data sets, model experimentation and validation, and production deployment.

• Technical lead of MLOps initiative to standardize and scale data storage, feature engineering, model training and experiment tracking.• Designed and implemented data models in Redshift and built ML experimentation workflow with Github Actions, Airflow, MLflow, Ray and Kubernetes.• Built classification and survival analysis models to predict progression of Cognitive Decline.

• Built code crosswalk database tool that connects common healthcare code systems in the US through hierarchical and synonymous relationships, allowing for translation and interoperability between data sources.• Built system to system API integration with external vendor to query Epic, Veradigm, Cerner and HIE data.• Built components of custom internal feature store library to make feature engineering more scalable across team by centralizing storage, processing and access to machine learning model features.• Built and deployed dozens of early detection machine learning models to predict onset and progression of diseases, leveraging Amazon Redshift to process billions of rows of data and MLFlow to track experimentation results.

• In charge of the data strategy for the company and hiring/building the data team• Managed a front end developer, data analyst, data engineer, and data scientist to build a suite of data products for device manufacturers and surgeons• Developed Epic App Orchard EHR integrations for 2 hospital systems• Built an event driven system to integrate legacy data and EHRs (HL7 or API) into one interoperable FHIR-based data warehouse using AWS Kinesis and DynamoDB

• Interacted directly with orthopedic surgeons, ambulatory surgical centers, and medical device companies to determine how to use data to address their needs and to demo new data-driven products• Performed ad-hoc analyses and created visualizations using Python and SQL to inform company strategy and to address specific needs of customers and partners• Built a Flask API backend for a patient-facing survey product to aid surgeons in making decisions for patients before and after surgery. Achieved double the industry standard for response rates• Built a statistical patient prioritization algorithm that allowed surgeons to determine which patients needed surgery the most based on customizable criteria and weights• Built a full stack manual data entry web app to increase speed and accuracy of data labelling for machine learning using HTML, CSS, JQuery, Flask and SQL

• Analyzed Medicare BPCI data and created visualizations to communicate insights and recommendations for specific surgeons• Built models for risk stratification of patients

• A member of the Golden One Center Medical Team• Provided emergency medical care for patrons, employees and professional athletes at NBA games and concerts

Developed several Data Science projects as part of a highly selective 3-month immersive program, utilizing various machine learning and statistical modeling methodsPredicting A Healthcare Provider’s Cost to the U.S. Government • Created a regression model with an R squared of 0.92 to predict how much a healthcare provider will spend on Medicare patients every year• Acquired data from the Center for Medicare and Medicaid Services of 990,000 healthcare providers• Used Ordinary Least Squares with Ridge and LASSO regularizationPredicting Lyme Disease Epidemic • Built a climate-based classification model with a ROC AUC of 0.96 to predict which U.S. counties will have a high incidence of Lyme Disease, the fastest growing contagious disease in the world at the time• Engineered a dataset from scratch by merging Center for Disease Control data with National Oceanic and Atmospheric Administration climate data which was parsed from 78,000 CSV files• Used K-Nearest Neighbors, Logistic Regression, Support Vector Machines and Random Forest algorithms optimized with grid searchA Divinely Inspired Text Generator • Built a Flask app with generative text model that can output divinely inspired religious sermons based on the King James Bible, the Quran, and the Torah• Used NLP techniques to do Sentiment Analysis using Vader and Textblob and to create LSTM Recurrent Neural Nets with Keras and TensorflowPredicting Emergency Medical Service Demand • Built 10 location-based time series models for each fire battalion in San Fransisco to predict how many ambulances will be needed every 30 minutes• Acquired 20 years of fire department data from the San Francisco Open Data Portal.• Used SARIMA, Croston TSB, and LSTM recurrent neural networks

• Worked on the Human Music Memory Map Project, studying how music-evoked autobiographical memories (MEAMs) affect well-being and pain perception in the geriatric population• In charge of data collection and user experience testing of the MEAMCentral app through in-person interviews with subjects • Designed web-based survey forms for subjects involved in the study

• Worked at fire stations 29 and 112 on their Advanced Life Support (ALS) ambulances, providing emergency medical care to citizens of Orangevale and North Highlands, CA• Part of an 192 hour pre-paramedic internship through American River College/UC Davis Fire Department

• Worked in the UC Davis Medical Center emergency room, performing tasks including EKGs, vital sign checks, CPR, bleeding control, spinal immobilization, and patient transportation• Part of an 192 hour pre-paramedic internship through American River College/UC Davis Fire Department

• Provided emergency medical care to patients at various events including construction sites, factories, concerts, raves, night clubs, tv show sets, and sporting events• Used ePCR system to collect patient data