Job Description

• Position: ​

​ ​ ​1) Hanwha Life: ​AI ​Modeling Research ​& Application

​ ​2) ​ Hanwha ​General ​Insurance: ​AI Modeling Research ​& ​Engineer

*Final candidates will ​be ​assigned ​to one of ​the above ​positions ​ based ​on the ​interview ​

​ ​ results.

• Position Type: Experiential Track (Full-time offer not available)
• Location: Yeouido, Seoul

• Responsibilities

1) [Hanwha Life]: AI Modeling Research & Application

- LLM Agent System Development

• Support planning and development of LLM-based agent systems tailored to domain-specific workflows
• Participate in research, development, testing, and refinement of AI agent systems

- LLM Training Data & Evaluation

• Support creation and evaluation of training and evaluation data using generative models
• Assist with data research, collection, preprocessing, and model evaluation

- AI Service Development

• Support development and service deployment using multiple AI models integrated with internal data and systems

2) [Hanwha General Insurance]: AI Modeling Research & Engineer

- AI Research

• Conduct research on core AI technologies, including machine learning and generative AI
• Design and validate new model architectures and algorithms
• Research and apply emerging AI technologies

- AI Model Engineering

• Develop conversational AI services based on generative AI models
• Design, develop, and optimize machine learning and deep learning models

• Required/Preferred Qualifications

[Hanwha Life]: AI Modeling Research & Application

[Required]

- Language requirements : Fluent in Korean and English (written and spoken)

- Experience training large-scale models using frameworks such as PyTorch, TensorFlow, or JAX

   - Familiarity with open-source model distribution platforms (e.g. Hugging Face)experience

   - Experience with or understanding of agentic AI frameworks and environments

   - Currently enrolled in a related field

[Preferred]

   - Experience in Korean sLLM research or product development

   - Authorship of papers at top NLP/ML conferences (e.g. ACL, EMNLP, ICML)

   - Experience building applications with Flask, FastAPI, Streamlit, or Gradio

[Hanwha General Insurance]: AI Modeling Research & Engineer

[Required]

- Language requirements : Fluent in Korean and English (written and spoken)

- Bachelor’s degree (or expected) in AI, Computer Science, Mathematics, or a related field

   - Solid understanding of machine learning and generative AI, with hands-on development

experience

[Preferred]

   - Experience publishing research papers or completing research-level projects

   - Experience developing conversational agents, such as AI assistants or chatbots

• Required Documents for Application

- Resume

- Case Study answer

※ You may submit your case study in either Korean

Case Study

• Q1과 Q2 중 하나를 선택하여 답변해주세요.

Q1. 암보험에 가입하려는 고객을 상담하는 챗봇을 생각해봅시다.

1. 암보험을 가입하려는 고객이 문의할만 한 문장들을 카테고리화 해봅시다.

(어떤 보험이 있는지 묻을수도 있고, 보험료가 얼마인지 물을수도 있겠습니다, 또 챗봇의 말에 긍정할수도 있고 부정할수도 있습니다.)

2. LLM을 활용해서 해당 문장의 유형 분류하는 유형 분류기를 만들어봅시다.

예를 들어서 ""안녕하세요!""라는 질문이 들어오면 ""greet""이라는 카테고리를 출력하여야 합니다.

3. (Optional) LangGraph, LangChain등을 활용해서 해당 유형에 알맞은 메세지를 출력하는 암보험 상담 챗봇을 만들어봅시다."

-------------------------------------------------------

Q2. SLM(Small Language Model)을 이용한 사내 지식(규정, 상품, 업무매뉴얼 등) 기반 챗봇 서비스를 구축하려 합니다.

• 예상 트래픽은 일 평균 1만 건의 질의이며, 응답 시간은 5초 이내가 요구됩니다.

• 답변의 할루시네이션 현상(근거 없는 답변, 계산 실수 등)과 같은 문제는 없어야 하며, 서비스 런칭 후 답변의 품질 유지가 되어야 합니다.

위 조건 하에 다음에 대해 서술해 주세요:

1) 금융권 망분리 규제 및 제한된 리소스(GPU 자원, 개발 시간) 상황을 고려한 최적의 아키텍처 구성 전략

2) 발생 가능한 문제들의 원인 진단 방법과 해결 방안 및 각 방안의 트레이드오프

Logistics & Disclaimer

• This is a paid internship.
• Applicants are advised to thoroughly review their applications for accuracy and completeness before submission. We will not be responsible for any disadvantages in the selection process resulting from any errors or omissions.
• In the case of any fraudulent recruitment practices or suspicion of any unethical behavior, the offer of internship may be revoked.
• False information or failure to provide verifiable documentation when requested will result in the withdrawal of the internship offer.
• Individuals with registered disabilities and those eligible for veteran’s benefits will be given preferential treatment in accordance with relevant laws and internal regulations.
• All communication regarding the progress of your application and screening results will be conducted via email. Please ensure your email address is correctly entered.