Seongwoo Choi

• Working on deploying in-house generative AI software on GitLab.
• Engineered a Naive RAG application using Python, LangChain, and Azure OpenAI to extract information from PDF documents.
• Developed a comprehensive RAG pipeline performance monitoring system to track LLM metrics for optimization.
• Conducted comparative analysis of Cohere embedding models against LangGraph Agents and Azure OpenAI models.
• Developed and showcased MVPs of Custom Copilot Solutions using Microsoft Copilot Studio.
• Led a demonstration of Azure OpenAI Services architecture to senior executives from LG CNS and Microsoft Asia.

• Researched and implemented a new algorithm in Graph Neural Networks on the AVA Dataset using PyTorch Geometric.
• Analyzed benchmark performance of various cloud computer configurations and virtual machine instances.
• Performed GPU analysis on cloud gaming workloads and learned about GPU architecture.
• Created automation for performance analysis workloads using PowerShell scripting.

• Optimized, debugged, and migrated an OpenCL application to Data Parallel C++ (DPC++), increasing runtime performance by 2.5X.
• Debugged and optimized deep learning models in TensorFlow on a Linux Server, achieving up to 5x speedup in training time.

Blockchain Financial Transaction Web Application: Created a decentralized web application for money transfers using React, TypeScript, Google Firebase, and Cardano APIs for security and transparency.

NVIDIA CUDA / OpenCL to Data Parallel C++ Migration Project: Migrated NVIDIA CUDA code to Intel oneAPI DPC++ and improved performance using Intel libraries and VTune Profiler.

Deep Learning Multi-classification on Chest X-rays: Engineered features and trained deep learning models to predict potential diseases from chest x-ray follow-ups using TensorFlow.