Machine Learning Engineer Graduate (E-Commerce Supply Chain & Logistics - LLM/Agent) - 2027 Start (PhD)

Join the E-commerce Global Supply Chain and Logistics team at TikTok. We are building AI-native capabilities for global logistics, including logistics agents, address intelligence, context engineering, agent evaluation, and workflow automation for complex supply chain operations. This role is for candidates who want to apply LLMs, agents, reinforcement learning, retrieval systems, and software engineering to real logistics problems at global scale.

We are looking for talented individuals to join our team. As a graduate, you will get opportunities to pursue bold ideas, tackle complex challenges, and unlock limitless growth. Successful candidates must be able to commit to an onboarding date by the end of the year. Please state your availability and graduation date clearly in your resume.

Responsibilities: - Responsible for the development of large language models, agent systems, and related intelligent systems for the supply chain and logistics of the global E-Commerce business. - Build and land domain LLM model capabilities for e-commerce supply chain and logistics, covering continued pre-training / CPT, SFT, preference optimization, reinforcement learning such as GRPO / PPO, reward or judge model design, model compression, inference cost and latency optimization, and business applications such as address correction, trajectory prediction, logistics cost analysis, customer service semantic understanding, and root-cause analysis. - Develop multimodal and structured understanding capabilities for logistics and supply chain scenarios, focusing on unified modeling of text, numerical time series, events, product attributes, images, and documents to support scenario simulation, explainable prediction, and integration with existing forecasting or decision systems. - Build core agent capabilities for team and business workflows, including AutoResearch for new solution exploration, Harness-based task decomposition and execution loops, RAG and knowledge retrieval, context understanding, skill / tool use, evidence grounding, and Clone & Adapt workflows for reusing proven solutions across markets and logistics scenarios. - Design and improve agent architecture and engineering systems, including runtime orchestration, memory and state management, model / tool routing, permission-safe execution, observability, benchmark and Golden Set evaluation, badcase attribution, regression testing, online feedback loops, and continuous evolution mechanisms that improve context, skills, workflows, and model behavior over time.