Last updated
Last updated
Name : Choi Jae Hun
Email : diadiahun0902@gmail.com
LinkedIn:
Address: Seoul, Geomcheongu
I am currently working as a AI Researcher. I have a strong interest to followings:
Implementing hyper-personalized assistants using RAG, Agents, and Prompt Engineering.
Extracting hidden information from user instruction.
Exploring and incorporating linguistic features and characteristics into prompt design, applying them across various sub-modules of RAG.
Comming Soon..😊
Responsible for overall tasks in the RAG Parts of the Gen.AI Team.
Developed the Assistant Module parts from the Gen.AI Server Solution. - Nov. 2024 - Mar. 2025
Designed and implemented an Advanced RAG Pipeline optimized for the service environment.
Implementing and using Langflow, a no‑code–based pipeline, along with Langfuse, a tracking platform.
Performed prompt engineering necessary for the development of the RAG pipeline.
Implemented access control and security features by restricting Retriever document access based on user permissions.
Developed a parameter-optimization pipeline for each chunking strategy to achieve optimal retrieval performance.
Apply prior research on the most effective chunking strategies.
Apply prior research on evaluation process, result of chunking strategy.
Apply prior research on chunk configurations for the best possible retrieval results.
Apply prior research on Text2SQL and Text2Cypher for integrating legacy ERP/MES systems.
Apply prior research on generating RAG evaluation datasets and establishing evaluation metrics for domain-specific data.
PoC Project with HD미포조선. - July. 2024 - Sep. 2024
Proposed methods for utilizing document metadata in the design of a Advanced RAG-based Chatbot Assistant project for the shipbuilding domain.
Designed and implemented a 2-Stage Retriever method to accommodate the requirement of periodic partial updates to document content.
Proposed and applied a chunk augmentation approach, table serialization, to accurately retrieve the given table.
Conducted a retrieval evaluation based on user scenarios provided by the PoC client, achieving 95% recall.
Assisted in the development of client projects using Python and Java.
Identified and fixed bugs within the software to improve user satisfaction.
Bachelor's Degree in Computer Science, Hannam University, Daejeon : Graduated: Feb 2023
Programming Languages : Python
Library : Numpy, Pandas, Matplotlib
Frameworks : fastAPI, Lagnchain, Langgraph, LlamaIndex, Langflow
Database : Faiss, ChromaDB, Postgresql, ParadeDB, Milvus, Qdrant
Tools : Git, GitLab, Visual Studio Code, Docker, PyCharm, Slack, Notion
Langchain OpenTutorials Contributors : Jan 2025 ~ Feb 2025
LangChain Global Tutorial Contribution with followings:
Original Tutorial : JsonOutputParser
New Tutorial : Text2SQL
New Tutorial : Synthetic Dataset Generation
Certificate of completion obtained for the LangGraph course accredited by LangChain Academy. : Jan 2025
Langchain Documents Contributors : Jan 2024
Acknowledged an error in the example for RetryParser in the official documentation due to a version update of Langchain.
Submitted a PR request to change from pydantic import BaseModel, Field to from langchain_core.pydantic_v1 import BaseModel, Field.
The issue was raised in Langchain's official repository under DOC: Error in Retry Parser example documentation #16698.
Presented the Gen.AI Server Solution at the Hannover Messe event :
