• Lead a 7-member Agile team as Scrum Master, overseeing project lifecycle from development to QA.

• Spearhead a GenAI initiative utilizing Retrieval Augmented Generation (RAG) and Llama 3.1 to develop an AI- powered tool for Primary Care Physicians (PCPs), which intelligently identifies and highlights missing patient information critical for accurate diagnoses, significantly improving the quality and efficiency of patient care.

• Develop and deploy an AI-driven document analysis tool, where the system extracts, summarizes, and intelligently suggests next steps for patient care, demonstrating advanced NLP capabilities integrated into healthcare workflows.

• Automate Selenium-based Python 837P file generation and upload process for IEHP, HPSJ, and CDCR on a Linux server, handling daily data scraping and parsing of 270,000+ user records, reducing costs by 25%.

• Integrate Twilio and Flask servers to enhance patient communication with over 100,000 SMS updates and in-app chat interactions, ensuring timely engagement.

• Deploy Docker images as Cloud Run jobs on Google Cloud Platform (GCP) to enhance the scalability and performance of cloud-based applications.

• Perform Database Administrator (DBA) duties in the team, optimizing databases by creating table indexing and foreign keys, optimizing SQL queries, and enhancing backend data retrieval speed and system performance.

• Evaluated student assignments, and coordinated instructional support with professors and TAs.

• Streamlined the OpenAI API to programmatically extract research characteristics including host, ethnic group, and research methodology

• Leveraged Python to gather a dataset of 18083 research papers from the Web of Science

Employed data cleaning, analysis, and visualization techniques using Python to aggregate the research dataset into a usable format

• Discovered patterns in migration research, which indicated that English publications tend to focus more on Global North issues

• Utilized unsupervised machine learning with an LDA model to classify research papers into different topics and draw insights from the data

• Used hyperfine EEG data to distinguish the regions of the superior temporal gyrus that process and hear words through machine learning algorithms

• Encoded 2.7 million EEG data from 9 different brain channels

• Built an SVM model for multiclass learning to find the mechanism of the cortex’s classification function for voice input tags in 4 general categories

• Produced results showing individual trials of hyperfine neural responses can differentiate different voice input categories despite the mean responses being the same for all categories