Hi there 👋 I'm Minsing-Jin(Minseong-Jin) and interested in AI.
- 🍚 I am in Nomadamas, which is a non-commercial open-source foundation in the AI field.
- 🌱 I’m currently learning: Computer Science, LLM action in field.
- 📫 I have experience in building an AI open-source project, classical ML, and LLM applications.
- LLM application in various domains
- Computer-use
- Outrageously weird topic for an Ig Nobel Prize
- Reinforcement learning
- Text mining solution
- large-scale text analysis automation
- LLM-based Opinion Mining
- LookerStudio visualization
- Build a classical machine learning project (tabular)
- Build an open-source project from scratch
- Technical writing
- Unstructured data preprocessing using pandas
- Fine-tuning OpenAI models and serving them via Docker
-
Markr AI Inc.Korean-SAT-LLM-Leaderboard - leaderboard benchmarking 10 years of Korean CSAT (College Scholastic Ability Test) exams, developed by the reputable KICE (Korea Institute for Curriculum and Evaluation). -
NomaDamasHorangi-leaderboard - Collarborate building Horangi leaderboard with W&B. Evaluate llm hallucination in Korean and join Korean-SAT-LLM-Leaderboard to horangi leaderboard. -
NomaDamasKoHalluLens - A Korean adaptation of HalluLens. Includes dataset translation and modified benchmark code to evaluate llm Korean hallucination. -
NomaDamasRAGchain - Building advanced Document Q&A framework. Fastest way to build a DQA system with retrieval and LLMs. Implement Dataset Evaluators. -
Markr AI Inc.– CS agent observability. A conversational data analytics tool for customer service centers, utilizing LLMs for analysis and Looker Studio for data visualization. -
Markr AI Inc.[H-AVOCA PoC] - A prototype of a public opinion analyzer 'H-AVOCA' explored AI applications for large-scale research data processing in Hankuk Research Inc. Analyze large news data and visualize the data in looker studio to assist analysis workflow. -
Markr AI Inc.[GPT-based AI Tool for Marketing Research] - Build experiment process to fine-tune openAI llm, using product review data in various domain. -
Markr AI Inc.[HRC classical machine learning] - Train Korean media statistics regression model to achieve bussiness-level performance in Hankuk Research Inc. Build classical machine learning experiment from scratch. -
Rice-BobbClueHunter - Chrome extension to find the evidence from the cited webpage in llm's answers automatically.
- 2th Ralphthon host which Geoffrey Huntly participated as a judge
- facebookreasearch/hallulens
- wandb/llm-leaderboard-korean: Add Korean hallucination benchmark.
- AutoRAG: Modify docs, Add model
🎀 Competitions
DaconJob Care Recommendation Algoritm Competition 15/728 CodeDaconJeju traffic prediction Competition 251/712 CodeKagglePCEO AI Club competition - exam score prediction 1/72021 KHU software festival2021 KHU software festival pet preference prediction junior award