Chris Nishimura Email & Phone Number

San Francisco, CA, US

San Francisco, California, United States

• Post-processed 1.5 million unstructured and structured UCSF medical records(~16GB).Transformed data into parquet format and stored in AWS S3, enhancing accessibility and analysis efficiency.
• Leveraged transfer learning and efficiently fine-tuned UCSF BERT model using a quantized low-rank adaptation (QLoRA) to predict patient length of stay.
• Currently investigating techniques to improve patient length of stay prediction performance, comparing the use of document embeddings in a XGBoost model to ensembling UCSF BERT with XGBoost.
• Frequently communicated with UCSF Director of Clinical Informatics to scope research needs and present results.

• Identified key customer profiles through customer segmentation using clustering, timeseries, and regression analysis.from a dataset of 10.4 million customer transactions spanning 4+ years.
• Led 6 students in a capstone project, effectively prioritizing leads and closing 500 net units for associated FinTech company.

Master Of Science - Ms, Data Science

Relevant Coursework: Deep Learning, Large Language Models, MLOps, Machine Learning, Distributed Computing, Inferential Statistics, A/B.

Bachelor Of Science - Bs, Cognitive Science With A Specialization In Machine Learning