🛠 Tech Stack

💼 About This Role

You'll help adapt Liquid's foundation models for Japanese language and culture, focusing on data curation, fine-tuning, and evaluation. You'll directly shape model performance for enterprise use cases in a fast-paced AI company spun out of MIT.

🎯 What You'll Do

• Curate high-quality Japanese text, audio, and multimodal datasets
• Synthetically generate or augment Japanese training data
• Train and fine-tune language and vision models for Japanese use cases
• Implement evaluation frameworks and benchmark model quality on Japanese datasets

📋 Requirements

• Native-level Japanese language proficiency
• Experience with HuggingFace inference, vLLM, or cloud APIs
• Deep understanding of Japanese model evaluation landscape
• Knowledge of Japanese pre-training data sources

✨ Nice to Have

• Experience building LLM pipelines and agents
• Familiarity with enterprise-grade data compliance requirements

🎁 Benefits & Perks

• 🔬 Hands-on experience with state-of-the-art AI technology
• 🌏 Directly shape foundation model performance for a complex language
• 🚀 Collaborative, fast-paced environment