AI Engineer
Seoul, Korea · 5 days/week, on-site
Develop NLP/VLM-based parsing engines that turn unstructured shipping documents into structured data.
Position Detail
MarinaChain is a global maritime logistics AI startup that structures unstructured data from the shipping and shipbuilding industries to deliver carbon-regulation compliance and decision-making solutions. We're looking for a capable engineer to help bring the parsing technology behind ParseAI, MarinaChain's core engine, to world-class level.
Key Responsibilities
- Parsing unstructured shipping reports: improving NER (Named Entity Recognition) and IE (Information Extraction) models that precisely parse the various unstructured text formats and documents generated by ships.
- Building a structure-aware parsing engine: optimizing algorithms that convert PDF and Word (.docx) documents to Markdown (.md) while preserving layout and formatting (bold, italic, etc).
- Designing data pipelines: generating source-map.json, which maps parsed results back to their original locations, and developing the logic that connects it with the chunking strategy.
Qualifications
- ML/NLP experience: hands-on experience in machine learning modeling and natural language processing (NLP)
- VLM proficiency: experience extracting and tuning data using Vision-Language Models
- Development stack: strong Python skills and proficiency with PyTorch- or TensorFlow-based deep learning frameworks
- Data-building experience: experience building and managing high-quality training datasets using semi-supervised learning and tools such as DVC
- Global communication: regardless of nationality, English or Korean communication skills sufficient for technical collaboration and documentation
Preferred Qualifications
- Document processing: experience with OCR services such as AWS Textract, or layout-analysis skills
- LLM agents: experience developing AI agents using RAG (Retrieval-Augmented Generation) or the ReAct framework
- Shipping domain knowledge: understanding of vessel operations data or related regulations (EU ETS, FuelEU, etc)
- Large-scale processing: experience serving models and optimizing GPU resources in a Kubernetes (K8s) environment
Tech Stack & Tools
Benefits & Perks
- Flexible culture: a development environment that supports growth, with a horizontal communication culture
Hiring Process
- Document screening
- Technical interview
- Culture fit interview
- Final offer
For the document screening stage, please share a portfolio or work samples where possible so you can best represent yourself.
Tags
Category
AI & Engineering
Location
Seoul, Korea
#104 7F, Apro Square, 55 Seocho-daero 77-gil, Seocho-gu, Seoul
Work Type
5 days/week, on-site
Experience
1-8 years of experience
Deadline
Rolling Recruitment