LLM Research Engineer
Job Summary
The company is seeking a Research Engineer to lead data-centric design and evaluation loops for computer vision and AI models. This role focuses on improving model performance by optimizing data quality, annotation policies, and training feedback cycles. You will work on real-world industrial applications, ensuring AI systems are stable and accurate through robust data-centric frameworks.
Responsibilities
- Design data for object detection and segmentation models.
- Manage preprocessing and quality control for image/video data.
- Develop annotation policies and evaluation guidelines.
- Build feedback loops for training, evaluation, and improvement.
- Collaborate with teams to enhance model performance.
Required Skills
- Machine Learning/AI project experience.
- Strong Python programming skills.
- Data-centric model optimization expertise.
- Experience with CV metrics and evaluation design.
- Knowledge of LLMs, VLMs, or Multimodal models.
Job Details
■ Recruitment Background
The company provides the generation, preparation, and evaluation processes of training data for AI model development. With the sophistication of computer vision (CV) models and increasing data quality requirements for production deployment, the factor determining performance has shifted toward how effectively one can design data. We require a member who can continuously elevate CV model performance from a data-centric perspective.
■ Challenges in the Computer Vision Domain
Computer vision equips computers with cognitive abilities to identify, locate, and track objects. Key technical fields include:
- Image Classification
- Object Detection
- Segmentation
- Tracking
- Simulation (Data augmentation)
■ Responsibilities
- Data design for object detection and segmentation models
- Preprocessing and quality control of image and video data
- Design of annotation policies and guidelines
- Design of model evaluation metrics and evaluation analysis
- Construction of a feedback loop spanning training → evaluation → data improvement
- Performance improvement through collaboration with the model development team
■ Core Focus
- Data design and quality control for CV models
- Construction of model evaluation and data improvement loops
- Image data processing and analysis using Python
- Model performance improvement from a data-centric perspective
■ Job Requirements
Required:
- Experience involved in data preprocessing, analysis, and evaluation design within machine learning or AI-related projects
- Experience in data processing and analysis using Python
- Experience addressing and improving model performance from a data-centric perspective
- Interest in AI quality, safety, bias, and evaluation metrics
Preferred:
- Experience designing data structures for LLM, VLM, and image/video models
- Experience in annotation work, including guideline design and quality standards
- Experience interpreting evaluation results using metrics such as F1 score and BLEU
- Experience building data pipelines and RLHF infrastructure
- Track record of research contributions or academic publications
- Development experience using generative AI (LLMs, agents, RAG, multimodal understanding)
- Expertise in robotics or physical AI data creation