Machine Learning Engineer, AI Agent

Our team focuses on the R&D of algorithm for TikTok international advertising customer growth. We leverage deep learning and large language model technologies to build an algorithmic system for advertiser acquisition and retention, optimize the operational efficiency of the international commerce platform, and empower the business to achieve sustainable long-term revenue growth.

What You'll Do

• Build the core "brain" of an AI agent for the advertising sales domain and track cutting-edge technologies. Explore mainstream Agent frameworks, continuously monitor the latest advancements in the AI Agent field, and introduce and validate the feasibility of new technologies—with a focus on enhancing automated task planning and scheduling, tool invocation, multimodal interaction, and deep research capabilities. Establish an Agent evaluation mechanism for Harness-driven iteration.
• Develop a Skills toolset for advertising sales and customer acquisition scenarios, building and optimizing a suite of high-performance, reusable Skills.
• Explore Agent automation and self-evolution, enabling the Agent to learn from real-world interaction data and continuously improve its decision-making and execution capabilities.
• Build stable and reliable Agent infrastructure, including performance and stability optimization of the Agent system.

Qualifications

Minimum Qualifications

• Master's degree or higher in Computer Science, Artificial Intelligence, Natural Language Processing, or related fields.
• Familiarity with at least one mainstream Agent framework (e.g., LangGraph, OpenClaw, Hermes, Codex, Claude Code, etc.) with hands-on experience.
• Deep understanding of the core principles of Large Language Models (LLMs) and AI Agents. Familiarity with mainstream Agent architectures (e.g., ReAct/PlanAct), Multi-Agent systems, and concepts such as Context Engineering and Memory.
• Solid AI/ML background with in-depth knowledge of post-training techniques such as SFT and RLHF. Practical experience in fine-tuning model capabilities for specific tasks is preferred.
• Excellent coding skills with strong fundamentals in data structures and algorithms. Proficiency in C/C++ or Python.
• Strong teamwork, collaboration, and communication skills.