Large Model Algorithm Researcher (Multimodal & Code AI) - Soaring Star Talent Program
Singapore · Full Time
Be the first to apply
- Experience
- Any
- Salary
- —
- Openings
- 1
- Posted
- 1 day ago
- Work mode
- In office
- Education
- PhD
- Eligibility
- Open to candidates holding a PhD with the requisite skills and experience in machine learning, large language models, computer vision, or natural language processing.
- Resume
- Required to apply
Where you'll work
Sign in to tell us what does and doesn't work for you here — it sharpens every match we show you.
Job description
About the Team and Project
The AI Innovation Center focuses on advanced AI infrastructure and pioneering research in areas like large language models (LLMs) and multimodal foundation models. The aim is to develop AI systems capable of understanding multiple languages and vast video content, enhancing user content consumption experiences. In parallel, the Code AI domain leverages LLMs to improve code comprehension and reasoning, thereby boosting programming performance and accelerating research and development efficiency.
One of the center's key projects in 2024 is VFM V1, a multimodal large model specifically designed for TikTok's commercial applications. This model matches top open-source benchmarks and surpasses other foundational models on business-oriented tests. The ongoing goal is to evolve these foundational models to integrate efficient perception and reasoning abilities, especially for multilingual and large-scale video content processing.
Key Challenges
- Improve the multimodal perception encoder beyond using a fixed frame rate by exploring adaptive frame rate strategies while integrating other modalities like audio and user behavior data.
- Develop methods to effectively fuse multimodal perception with cognitive processing to enhance the comprehensive understanding capabilities of the model.
Qualifications
- PhD degree holder, preferably with publications in machine learning, computer vision, or natural language processing.
- Strong programming capabilities and solid understanding of data structures and algorithms, proficient in C/C++ or Python. Candidates with accolades from coding competitions (ACM/ICPC, NOI/IOI, Top Coder, Kaggle) will be prioritized.
- Research expertise in machine learning focusing on large language models and generative AI technologies.
- A passion for technology, excellent analytical and problem-solving skills, enthusiasm for tackling complex issues, strong communication skills, and a collaborative team spirit.
About ByteDance
Founded in 2012, ByteDance aims to inspire creativity and enrich life with products like TikTok, Lemon8, CapCut, and others. The company values innovation, diversity, and inclusion, fostering an environment where employees' unique perspectives contribute to global connectivity and content creation. ByteDance encourages a culture of curiosity, humility, and continuous improvement to generate meaningful impact for users and communities worldwide.
Minimum education
Doctorate