Lav P Patel

Architected and built an AI-powered clinical documentation platform CareScriptMD that converts real-time patient visit audio into structured medical notes for memory clinic providers, eliminating hours of manual charting per clinician per day. • Designed and shipped a full-stack clinical AI system end-to-end: Web application with real-time audio capture, OpenAI Whisper speech-to-text pipeline deployed as a custom MLflow serving endpoint on Databricks, and multi-model LLM orchestration (Claude, Llama) for structured note generation, all within HIPAA-compliant infrastructure. • Engineered a production ML pipeline on Databricks: registered Whisper as a custom PyFunc model in MLflow/Unity Catalog, built auto-scaling GPU serving endpoints with scale-to-zero economics, and implemented service principal authentication, reducing inference costs by designing a hybrid model routing strategy (Claude for accuracy-critical notes, Llama for high-volume intake). • Navigated dual-track institutional compliance: secured IRB approval as a Quality Improvement initiative, started security reviews across both university (KUMC) and hospital systems simultaneously, and implemented CAS 2FA enterprise authentication, clearing the regulatory path for clinical deployment. • Operated as sole technical builder on a cross-functional team with the PI (Jeff Burns, MD), neuropsychologists, and clinical coordinators, translating ambiguous clinical needs into ready to ship product.

* Lead Software Engineer for AllofUs Heartland consortium. Develop Observational Medical Outcomes Partnership (OMOP) from scratch and also make code available for other sites in the Heartland consortium (University of Missouri, University of IOWA and UNMC) · Successfully led a cross-functional team of 10 engineers in the development and implementation of a cutting-edge biomedical informatics software system. · Architected and oversaw the construction of a COVID-19 data pipeline at KUMC, connecting approximately 10 prestigious hospitals in just 7 days, a task that typically takes 1.5 months to accomplish. Utilized Python, Java, Spark, Databricks , DBT, Airflow · Headed the Knowledge Discovery Team at KUMC and authored 20 papers on COVID-19. · Witnessing the immediate real-world impact of my skills made this experience exceptionally rewarding · Employed Python, PyTorch, Scikit-learn, Triton, CUDA, C++, and SQL

· Significantly enhanced the execution speed of PCORNET Common Data Model (CDM) task, resulting in a remarkable reduction from 233 hours to 5 hours (47X faster). Task was funded by CDC, NIH, FDA, and PCORI · Leveraged Python, Java, Spark, SQL, Airflow, Bash, and Jenkins

·Improvedspeedandflexibilityofaslowdatapipelineforadatasource,reducingprocessingtimefrom4hoursto5minutes by re-architecting the system. Utilized Python, Spark, SQL, Jenkins, and Bash during the project · Debugged and optimized the main Extract Transform and Load (ETL) process for thedepartment, decreasing processing time from 28 days to 7 days by identifying bottlenecks and implementing fixes. Utilized Python, pandas, SQL, Jenkins, and Bash in the process · Debugged an open-source Java software called I2B2, optimizing the system to achieve query completion within seconds, compared to previous instances where queries never finished. Utilized Java and SQL in the debugging process

· Led the initiative to ensure the secure storage and protection of over 21 Million Medicare and Medicaid patients' data, adhering to industry standards such as NIST-850, HIPAA, CIS benchmark, and utilizing vulnerability scanners, Jenkins audit jobs. · Maintained an impeccable record of zero security breaches to date. · Successfullyimplementedahighlycomplexsoftwaresolutionwithina15-daytimeframe,enablingfederatedqueriesacross 50+ hospitals and delivering instantaneous results. · Used Infrastructure as Code (Iac : Terraform, Azure, Docker, K8s, Ansible, Jenkins) the aforementioned software for enhanced reproducibility and scalability. · Actedasafront-endengineer,developingimpactfulwebpages.UtilizedadiversetechnologystackincludingJava,Tomcat, Wildfly, Apache, Oracle, MySQL, JavaScript, HTML, and CSS.

· Implemented a reimagined import process using Distributed Systems (Spark invoked through Terraform on Azure) to significantly improve the speed of the hospital patients' data loading process, reducing the time from 2 days to just 4 hours. · Successfully installed REDCap (PHP, Apache, MySQL, Linux, K8s, Terraform, Azure) while proficiently developing custom add-on software using PHP. This enhanced functionality streamlined the evaluation process for hospital residents, eliminating the need for manual methods and reducing the timeframe from a week to instantaneous results · DevelopedcomprehensiveguidelinesandspearheadedthecreationofgoldenOSimages,resultinginadrasticreductionin server building time from a one-month project to a streamlined 10-day process. Collaborated effectively with stakeholders from multiple teams, including Our Team, Security Team, and Data Center Team. Used bash, Ansible, Jenkins, · ServedastheDatabaseAdministrator(DBA)forthedepartment,proficientlymanagingtaskssuchasgrantingpermissions, creating roles, updating database links, and onboarding users. Resolved performance issues to ensure optimal database functionality · Demonstrated expertise in creating new servers and implementing robust security measures based on industry standards such as HIPAA and CIS benchmark. Utilized Python, Bash, Ansible and Linux in alignment with best practices · Expertly debugged and resolved servers and systems performance issues, ensuring seamless operations and maximizing efficiency.