Research Scientist - Large-Scale Machine Learning Systems (SysML) - Global Frontier Tech Recruitment Program - 2027 Start (PhD)

Team Overview

The Applied Machine Learning (AML) team focuses on pioneering research and deployment of cutting-edge machine learning technologies, including large pre-trained models and device-cloud collaborative learning. Their work spans several application areas such as search, recommendation, advertising, auditing, and federated learning. The team has a robust foundation in scientific research combined with practical engineering and product development. Team members have expertise in NLP, computer vision, multimodal learning, graph computing, and more, with over 100 publications in top academic conferences.

Role and Opportunities

We seek talented individuals to join our team starting in 2027. As a new graduate, you will have the opportunity to pursue innovative ideas, address tough challenges, and experience expansive personal growth. The position requires candidates to confirm availability to onboard by end of 2027 and clearly state graduation date in their applications.

Research Focus - Large-scale Multimodal Recommendation Systems

Large-scale recommendation systems are increasingly applied across various product lines involving short-video, text-based multimodal, and image platforms. Modality-specific data is playing an ever-growing role in improving recommendations. Recent breakthroughs and industry insights indicate that integrating modality information as generalized features can significantly support recommendation engines and other business applications.

Challenges

• Complexity in unifying multimodal data representations and performing efficient fusion.
• High computational requirements for training and inference of extremely large models.
• Balancing system stability and efficiency when handling long sequence modeling in end-to-end pipelines.
• Algorithm and engineering co-design complexity combined with adapting to heterogeneous hardware.

Impact and Value

• Technical impact: Enable breakthroughs in fusing multimodal representations and alleviating training/inference bottlenecks for ultra-large models. Enhance co-design methodologies between algorithms and engineering, and advance support for diverse hardware platforms including domestic high-performance frameworks.
• Business impact: Improve recommendation accuracy and generalizability in multimodal contexts, surpass current modality constraints, empower multiple products such as short video and community platforms, decrease computational overhead, and drive scalable business growth.

Candidate Qualifications

• Currently pursuing or recently completed a PhD in Artificial Intelligence, Computer Science, Computer Engineering, or closely related fields.
• Strong programming skills in at least one language such as C, C++, Go, Python, or Java operating in a Linux environment.
• Deep understanding of distributed system design principles, with hands-on experience in building and maintaining large-scale distributed systems.

Preferred Experience

• Knowledgeable about Kubernetes architecture and experienced in cloud-native system development.
• Hands-on with at least one major machine learning framework such as TensorFlow, PyTorch, or MXNet.
• Experience with backend development using technologies like Django or Flask.
• Background in areas like AI infrastructure, hardware/software co-design, high-performance computing, ML hardware architectures (GPUs, accelerators, networking), machine learning frameworks, ML for systems, or distributed storage.

About the Employer

ByteDance was founded in 2012 with the goal of inspiring creativity and enriching life. Operating globally, the company offers more than a dozen products including TikTok, Lemon8, CapCut, and Pico, plus region-specific platforms like Toutiao, Douyin, and Xigua, making content consumption and creation more accessible and engaging.

Why Work Here

Passion for creativity drives the company. ByteDance nurtures a collaborative, diverse, and innovative culture, empowering people to express themselves and connect meaningfully. Adopting an “Always Day 1” mentality, the team strives for significant innovation and impact in a rapidly growing environment.

Diversity & Inclusion

The organization is committed to fostering an inclusive workplace where diverse skills, experiences, and perspectives are valued. Reflecting its global user base, ByteDance embraces varied voices and backgrounds to foster creativity and enrich lives.