MarqVision
Foundation Model Engineer
Seoul, KoreaPosted today
What you'd do
- Design and train domain-specific foundation models for online product images, text, and IP data.
- Lead training data pipelines through model evaluation and tie performance to product metrics.
- Deploy trained models to production and integrate across multiple AI pipelines.
What they want
- 2+ years of development experience with LLMs or multimodal models.
- End-to-end experience operating models from training through serving.
- Integration design experience in environments with linked AI pipelines.
Nice to have
- Domain-specific VLM pre-training and fine-tuning experience.
- GPU cluster distributed training and serving optimization.
- Ability to collaborate in English.