Prem Sagar Kumar

I worked on the ICD project, which was part of ICA, for 45 days as a GEN AI Engineer. I was given access to Kubeflow and understood the workflow of the ICD project. The pipeline was already created, and I executed it with different types of YAML configurations based on schema. The schema was related to various modalities and indications. I made changes in three YAML files based on modality, indication, and prompts. Additionally, I wrote a Python script for a post-processing function.

The main objective of this project was to generate GPT files, post_processing_gpt, and rollup files, and to get them reviewed by clinicians. After the review, the guidelines needed to be updated in the UI. I generated the initial three files, but updating the UI was a bit slow because reviewing the files was also a major task for the clinicians.

I worked on a POC for BO membership, which was part of the Innovation Council at Carleon, for almost four months. During this time, I learned about some LLM models like RAG and LLaMA. The main goal of this project was to automate the mainframe screen. I automated error handling using Python scripts and generated different file formats for various zones in the USA. Additionally, I fetched data from APIs, which was in JSON format, and wrote Python scripts to populate different fields with different data.