AI Engineer Intern

Overview

This role centers on supporting Tencent's international gaming operations by developing and optimizing AI computing infrastructure focused on large language models (LLM) and reinforcement learning frameworks.

Key Responsibilities

• Design and implement large-scale distributed training systems, including data and model parallelism techniques such as Tensor Parallelism, Pipeline Parallelism, and ZeRO optimization to maximize GPU utilization and maintain stability of extensive training tasks.
• Develop and refine intelligent compute scheduling strategies to overcome resource constraints in complex gaming environments by leveraging precise resource management, fault-tolerance processes, and efficient checkpoint mechanisms.
• Handle the entire engineering pipeline from training AI models to deploying inference services, contributing to tasks like operator profiling, model quantization, and building high-throughput inference workflows that accelerate AI integration in gaming products.
• Utilize AI-assisted coding tools to enhance productivity and lead engineering excellence practices including automated testing and infrastructure governance to ensure robust and dependable system operation.

Required Qualifications

• Undergraduate degree or higher, preferably in Computer Science, Computer Architecture, High-Performance Computing, or related disciplines.
• Strong programming skills in Python, C++, or Go and solid knowledge of the PyTorch framework.
• Experience with distributed training platforms such as DeepSpeed, Megatron-LM, or similar technologies.
• Good grasp of distributed system concepts; familiarity with technologies like NCCL, RDMA networking, or high-performance storage solutions and container environments (Docker, Kubernetes) is advantageous.
• Prior exposure to AI coding assistants like GitHub Copilot or Cursor is a significant benefit; experience in engineering governance, automated benchmarking, or stress testing is highly desirable.
• Excellent problem-solving abilities, logical reasoning, and collaboration skills within multidisciplinary technical teams; fluency in English communication is required.
• Additional advantage for candidates experienced in backend high-performance architecture or hands-on LLM training and inference development.