AI Infra Optimization Engineer Graduate – 2027 Start

About the role

Join TikTok's fast‑growing E‑commerce Recommendation Infra team as an AI Infra Optimization Engineer Graduate. You will work on cutting‑edge large language model (LLM) and recommendation technologies, helping to build and scale the next generation of recommendation infrastructure for live‑streaming, short‑video, and product recommendation services.

Key responsibilities

• Integrate LLM and recommendation technologies, developing and deploying training and inference pipelines for foundation models.
• Design and implement high‑performance model infrastructure using open‑source components such as Vllm, SGlang, Megatron, and VeRL.
• Collaborate with algorithm teams to research, prototype, and productionize end‑to‑end generative recommendation solutions.
• Re‑architect recommendation models and the full training/inference stack, focusing on ultra‑long‑context user behavior modeling and KV‑cache sharing across sessions.
• Optimize GPU kernels and multi‑GPU parallelism for extreme throughput and performance, employing automated, agentic optimization paradigms.

Required profile

• Recent graduate (or graduating by the end of the year) with a strong interest in AI infrastructure and recommendation systems.
• Ability to commit to an onboarding date by the end of the calendar year.
• Passion for tackling complex engineering challenges and delivering production‑ready solutions.

Required skills

• Proficiency in Python and deep‑learning frameworks, especially PyTorch.
• Experience or familiarity with LLM components such as Vllm, SGlang, Megatron, or VeRL.
• Knowledge of GPU programming, multi‑GPU parallelism, and performance optimization techniques.
• Understanding of recommendation system architectures and large‑scale model training/inference.