Bhishan Poudel, Ph.D. Physics

I worked as a Data Scientist II in Healthcare company called Cencora . During the time, I worked on a project called "Chatbot for Benefit Verification" where I created a Plugin for ChatGPT alike inhouse tool called "CenChat" using Generative AI tools OpenAI, Langchain with with help of custom prompt and Few-Shot examples, then deployed the app as Azure Service App by hosting the source code in Azure Git and storing data at Azure Storage Blob. This chatbot is capable of question answering patient service questions like "how many Pfizer Bridge patients completed benefit verication in 2024" and so on. Another project I worked on is, "Regeneron Drugs Eylea to EyleaHD Transition", in this project I did a detailed study of eye patients transition to and from Eylea to EyleaHD using Databricks SQL and presented a PowerBI report to the stakeholders. I also calculated average shipment dates to the patients who transition from PAP to Copay. Another project I worked on is, "Timeseries Predictions for Selected Drugs", in this project, I used various statistical models and advanced time series models such as ARIMA, Prophet, and greykite to predict the number of sales of given drugs after one quarter, so that we can plan ahead and do the preparations. Overall, having worked on these advanced projects, I have expert level skills in Generative AI Large Language Models (LLM), OpenAI Embeddings, Creating Chatbos and Document Parsers using Langchain and LlamaIndex, along with traditional machine learning models such as scikit-learn and xgboost and time series modelling methods like ARIMA and facebook prophet.

Education wise, I did my Ph.D. in Physics and Masters in Physics from Ohio University, USA. During my study time, I was heavily interested in Machine Learning and Deep Learning and took various courses in Computer Science Department as well as online courses from Coursera and Udemy. I invested significant time in data analyis using pandas and answered thousands of python/pandas related question in stackoverflow to help fellow developers gaining top level contribution award. In my PhD research, my research topic was determining the ellipticity of the galaxies using various machine learning and deep learning techniques. This Ph.D. research helped me learn a plethora of statistical concepts, visualization tools, machine learning and deep-learning skills. When I was doing my Masters Degree in 2014, I designed and maintained astronomical image processing tool called jedisim which helped me learn the concepts of OOP, GitHub version control, continuous integration, unit testing, code debugging, documentation (Sphinx and ReadTheDocs) and various other software development skills.