Incheon, South Korea | Full-time | On-site / Hybrid
About the Role:
We are looking for a Middle / Strong Middle NLP / LLM Engineer to develop production AI solutions across RAG, LLM applications, AI assistants, document processing and multilingual NLP, working closely with senior engineers.
Key Responsibilities:
- Develop and improve RAG applications: chunking, embeddings, retrieval and reranking.
- Build LLM workflows and agents with LangChain, LangGraph, LlamaIndex, CrewAI or similar.
- Develop query rewriting, intent classification, routing and structured extraction.
- Integrate LLM APIs/models into production systems.
- Work with OpenAI, Gemini, Hugging Face and PyTorch.
- Contribute to SFT, LoRA, QLoRA, PEFT and LLM evaluation.
- Work with OCR, PDF/DOCX, STT and conversational memory.
- Build APIs with Python / FastAPI and work with Docker, Git, testing and CI/CD.
- Monitor and improve model quality, latency and cost.
- Take ownership of individual features/components with senior-engineer mentorship.
Requirements:
- 2–4 years in NLP / ML / Information Retrieval / Applied AI.
- Strong Python skills.
- Hands-on experience with LLM / RAG / AI-agent applications.
- Understanding of transformers, embeddings, prompting, retrieval, chunking and reranking.
- Experience with LangChain / LangGraph / LlamaIndex / CrewAI.
- Experience with Qdrant / Pinecone / Milvus / Weaviate / Elasticsearch.
- Experience with OpenAI / Gemini / Hugging Face and familiarity with PyTorch.
- Understanding of LLM evaluation and hallucination mitigation.
- FastAPI, REST, Docker, Git.
- Multilingual NLP is a plus.
- Professional English.
Nice to Have:
- SFT / LoRA / QLoRA / PEFT
- BM25 / RRF
- RAGAS / DeepEval / LangSmith / Langfuse
- OCR/STT
- Graph RAG
- vLLM/TGI/Triton
- GPU optimization
- Korean is a plus.
Visa and relocation support may be available.
