[AI Research Div.] Research Scientist - Specialized Model (2년 이상 / 인턴)
Salary not listed • Seoul; Seoul, South Korea • Internship
Posted 1mo ago
Remote Work Policy
On-site
Employment Type
Internship
Categories
LLM Engineer
About the job
We are seeking a Research Scientist to specialize in developing tailored AI models for non-gaming domains. This role focuses on creating "Specialized Models" that offer superior quality and cost-efficiency compared to general-purpose frontier model APIs. You will be instrumental in adapting open-source foundation models for specific tasks through fine-tuning, evaluating their performance against established benchmarks, and optimizing them for both quality and cost. A key aspect of this position involves designing and implementing robust evaluation frameworks to ensure the reliability and effectiveness of these specialized models. The work encompasses the entire AI/ML model development lifecycle, from data curation and model training to evaluation and automation, aiming to build end-to-end development pipelines that deliver tangible business value.
Responsibilities
- Develop specialized AI models for non-gaming domains.
- Evaluate frontier and open-source models based on quality and cost.
- Establish quality and cost baselines for models.
- Improve target task performance using distillation and SFT.
- Curate and refine training data, design experiments, and analyze errors.
- Optimize models and training strategies considering both quality and inference cost.
- Design and operate evaluation systems for specialized models.
- Build evaluation protocols including quantitative metrics, LLM-as-a-Judge, and pairwise comparisons.
- Analyze failure cases and model defects using quantitative and qualitative methods.
- Determine deployment readiness and improvement priorities based on evaluation results.
- Build end-to-end development pipelines from data to evaluation and improvement.
- Expand the specialized model development framework to new clients and domains.
- Create practical business value in terms of model quality and cost.
Requirements
- Degree in Computer Science, Machine Learning, Artificial Intelligence, Data Science, or equivalent professional experience.
- Experience developing and improving AI/ML models using deep learning frameworks like PyTorch.
- Proven experience in meaningfully improving model performance through data analysis, training strategy, hyperparameter tuning, and error analysis.
- Experience with post-training LLMs for specific tasks.
- Experience applying post-training techniques (SFT, distillation, DPO/RLHF) or fine-tuning open-source LLMs, including training data curation and defect analysis.
- Experience designing and operating evaluation systems for LLM-based systems.
- Experience with quantitative metrics, LLM-as-a-Judge, pairwise comparisons, statistical interpretation of results, and building improvement loops.
- Problem-solving skills that consider data, models, and evaluation comprehensively.
- Experience defining and solving problems in real-world service environments, considering both quality and cost, based on an understanding that model performance depends on data quality, configuration, and evaluation reliability.
- Ability to travel internationally without restrictions.