ByteDance Seed 2027 Foundation Model Campus recruitment is now open (including internships)
We have officially started recruiting for the Seed Foundation Model Campus for 2027 graduates.
We firmly believe that true technological progress comes from continuous efforts to reach new heights. Amid the immense opportunities of the AI era, the Seed team goes beyond iterating on models. Instead, we work on core cutting-edge technologies, drive breakthroughs in next-generation AI paradigms, and constantly explore the boundaries to advance the frontiers of intelligence.
Join today to be at the forefront of building cutting-edge AI capabilities and driving the next breakthrough.
For more information, click to view the project page
- new graduate (Applicable to undergraduate, master’s, and doctoral students graduating from September 2026 to August 2027)
- intern student (For students scheduled to graduate after September 2027)
Direction of technological research
basic model
- Pursuing the edge of the foundation model‘ Understand the limits of functionality and efficiency and continually optimize model architectures, training paradigms, and core functionality.
visual intelligence
- We will focus on building new paradigms for multimodal pre-training, multimodal fusion, integrated generation and understanding for visual inference, multimodal agents, and other related fields.
voice intelligence
- Explore integrated audio generation and understanding, personalized dialogue, speech agents, music generation, and other related areas.
machine learning system
- We will focus on key technical issues such as ultra-large-scale training, fundamental model reinforcement systems, inference optimization, hardware and software co-design, and compilation acceleration.
LLM application
- Optimize usability for key tasks, build memory and personalization systems, and integrate search and tool invocation to comprehensively improve the user experience across ByteDance applications.
embodied intelligence
- Explore generic robot foundation models, data algorithms, and reinforcement learning to continuously improve robots’ ability to perceive, make decisions, and execute in real-world environments.
What you can achieve here
Tackling AI’s toughest frontiers
- Join us in building industry-leading Seed Foundation and GenMedia models and explore the next generation of AI.
Fully supported access to R&D resources
- Our robust AI infrastructure gives you access to sufficient computing power, data, and engineering resources to focus on overcoming your core technical challenges.
Create real value for the world
- We have competitive advantages in both consumer and enterprise products. Your technological achievements provide tangible value to hundreds of millions of users.
Earn rewards commensurate with your technical contributions
- We offer a competitive compensation package. Qualified graduates and high-achieving interns will have the opportunity to participate in additional incentive programs designed for AI talent.
Deliver industry-leading results with us
learn more: https://seed.bytedance.com/en/models?view_from=homepage_tab
Employee C
Employee C, a recent graduate with less than a year of work experience, was deeply involved in the difficult task of building the Seedance 2.0 pre-training architecture, researching effective large-scale scaling algorithms, and achieving breakthroughs in this field.
Intern Y
During the research and development of Seed1.8 and Seed2.0, Intern Y adopted state-of-the-art RL algorithms to stably and efficiently integrate multidimensional expert features into the integrated base model, raising the bar for the overall performance of the model.
Employee Z
Employee Z, a new graduate with less than one year of working experience, was deeply involved in the research and development of Seed’s end-to-end simultaneous interpretation model, helped tackle the data cold start problem after training, and formed a “data-training-evaluation” closed loop to enhance the functionality and effectively improve the model’s performance in aspects such as translation quality, latency, and timbre.
Intern R
During his internship, intern R conducted in-depth research to explore ways to address high costs in model inference and RL simulation environments. The related research results have been accepted at major international system conferences and widely applied in Doubao online services and Seed foundation model post-training.
location of base
Beijing, Shanghai, Shenzhen, Hangzhou, Singapore, San Jose, Seattle
How to apply
Click to check out
Mainland China position
Positions outside mainland China
If you have any questions, please contact us.
Seedcampus@bytedance.com
