Graduate AI Infra Research Engineer

Job Description

Graduate AI Infra Research Engineer | ByteDance

The Tone:
This is an early career role for a Graduate AI Infra Research Engineer at ByteDance, located in the United States, with an onboarding date expected by the end of 2026. The Compute division at ByteDance focuses on building large-scale and highly available cloud infrastructure, which supports both public cloud products like VolcEngine ECS and ByteDance’s internal products, including TikTok. This specific position within the Compute-US team offers recent PhD graduates an exceptional opportunity to pursue bold ideas, tackle complex challenges, and drive the development and research of the compute infrastructure platform, specifically optimizing the AI Infra stack.

The TL;DR
• Role: Early Career
• Location: United States
• Pay: $136800–$259200 yearly
• Team: Compute-US team within the Compute division
• Mission: Develop and optimize the AI Infra stack, covering training, inference, and AI agents, to advance large-scale cloud infrastructure.

What You’ll Actually Do
• Development: Develop key technologies to optimize the AI Infra stack, focusing on training, inference, and AI agents.
• Collaboration: Collaborate with academic institutions and open-source communities on joint development initiatives.
• Analysis: Monitor the latest technological advancements from academia and industry, conducting deep-dive analyses.
• Publication: Present research findings and product innovations in academic papers.

The Must-Haves
• Background: Early Career Doctorate Degree (PhD) in Computer Science, Computer Engineering, or a closely related technical discipline is required for this research-focused engineering role.
• Experience: Candidates must possess demonstrated experience in at least one of the following highly specialized AI infrastructure domains:
• LLM training infrastructure, including advanced optimizations for post-training workloads such as reinforcement learning training or knowledge distillation techniques.
• LLM inference infrastructure, with a focus on performance improvements for inference engines, more efficient execution parallelism, or low-level GPU kernel optimizations.
• AI Agent infrastructure, covering aspects like computer-use agents, coding agents, agent memory management, or agent sandbox environments.
Furthermore, a commitment to proactive continuous learning, a demonstrated enthusiasm for cutting-edge AI technologies, and a strong ability to quickly grasp and apply new technologies are essential.
• Skills: Excellent communication and teamwork skills are crucial for collaborating effectively with internal teams, academia, and open-source communities.
• Bonus: Preferred qualifications include being a frequent contributor or maintainer in prominent AI Infra related open-source communities, such as vLLM or SGLang. Additionally, having publications in top-tier Computer Science conferences like OSDI or MLSys would be highly advantageous.