PhD Intern – GPU Infrastructure Management

Job Description

Research Intern (AML) – 2026 Start (PhD) | ByteDance

The Tone:
This is a PhD Internship at ByteDance. ByteDance, a company founded in 2012, builds a suite of over a dozen products, including TikTok and Douyin, focused on inspiring creativity and enriching life. This role is within the central Applied Machine Learning team, which delivers state-of-the-art solutions for the company’s recommendations, ads, and search systems. Interns actively contribute to ByteDance’s products, research, and its future plans, gaining hands-on experience and collaborating with industry experts.

The TL;DR
• Role: Internship
• Location: Unspecified
• Pay: $60 hourly
• Team: Applied Machine Learning team (central team)
• Mission: Optimize resource scheduling strategies to generate additional AI computing capacity and manage Group-wide GPU resources effectively.
• Tech Stack: Python, Golang, C++, MySQL, Redis, MQ, OpenTSDB, Prometheus, InfluxDB, ClickHouse

What You’ll Actually Do
• GPU Pool Construction: Construct and maintain the centralized GPU computing pool.
• Resource Scheduling: Optimize resource scheduling strategies to generate additional AI computing capacity.
• GPU Data Management: Collect, process, and mine GPU data across the entire Group.
• Resource Lifecycle Management: Leverage data-driven insights to manage Group-wide GPU resources, including budgeting, delivery, efficiency optimization, and secondary utilization.

The Must-Haves
• Background: Doctorate Degree. Student. Currently pursuing a PhD in Computer Science or a related field.
• Experience: Hands-on experience in the design, development, and maintenance of large-scale distributed systems, coupled with a solid understanding of distributed system principles.
• Skills: Proficiency in at least one programming language among Python, Golang, or C++ with good coding practices; proficiency in common storage and middleware systems such as MySQL, Redis, and MQ, including troubleshooting and performance tuning; familiar with common time-series data components like OpenTSDB, Prometheus, InfluxDB, and OLAP databases such as ClickHouse.
• Bonus: Possesses a strong sense of responsibility, excellent learning and communication skills, is highly self-motivated, and a good team player.