Conversational AI Large Model Researcher - Global E-commerce

Team Overview

The team is focused on developing a leading large-scale dialogue system supporting hundreds of millions of daily users within the Douyin e-commerce platform. This covers key e-commerce roles such as customer service for platform users, merchants, and influencers, as well as innovative smart shopping assistant applications. Through continuous innovation, the team delivers comprehensive intelligent dialogue solutions boosting both operational efficiency and user interaction quality.

Research Goals

The primary objective is to create a large language model (LLM)-powered customer service chatbot tailored for TikTok and Douyin's e-commerce ecosystem. This chatbot autonomously handles the entire query process, from clarifying requests to negotiating and executing solutions.

Rationale and Challenges

The powerful conversational and reasoning capabilities of LLMs make them especially suited to intelligent customer service, aiming to meet or surpass the quality of expert human agents. Key research challenges include:

• Developing a multi-agent system structure where the planning-agent identifies user needs and constraints; the reply-agent ensures responses comply with platform service policies and maintains seamless user negotiation; and the tool-agent manages tool legality and execution dependencies.
• Maintaining strict compliance with platform service guidelines such as refund policies and coupon restrictions.
• Facilitating dynamic, real-time adaptability of chatbot responses guided by user behavior and feedback, enabling personalized and evolving service that mimics human flexibility.
• Implementing self-reflective mechanisms for the LLM to monitor, analyze, and refine its own behavior, enhancing decision-making on complex or ambiguous cases.
• Handling complex image inputs including shipping orders, bank transactions, damaged goods images, and seller certifications to improve problem resolution effectiveness.

Qualifications

• Doctorate degree in computer science, artificial intelligence, software engineering, mathematics, or related disciplines.
• Strong foundational knowledge in machine learning, with deep expertise in deep learning, multimodal models, and generative models, complemented by excellent mathematical skills and self-learning capabilities.
• Proficiency in machine learning frameworks such as PyTorch and TensorFlow, along with strong software engineering and coding skills.
• Experience with multimodal large models, especially handling long-text processing or working within film/television domain applications, is desirable.
• Preference for candidates with published research in top-tier conferences/journals (e.g., NeurIPS, ICML, CVPR, ICCV, ECCV, IJCAI, AAAI, KDD, SIGIR, WWW, ACL, PAMI, IJCV) or demonstrated competition experience.

Company Overview

TikTok is a premier platform for short-form mobile video content, committed to inspiring creativity and delivering joy globally. Headquartered in Los Angeles and Singapore, with multiple international offices, TikTok fosters an innovative culture that encourages authentic self-expression and connection worldwide.

Why Work With Us

Collaborate with a diverse global team driven by curiosity, humility, and impactful innovation. We embrace challenges with resilience and maintain an "Always Day 1" philosophy, empowering continuous learning, creativity, and meaningful breakthroughs.

Diversity & Inclusion

TikTok is dedicated to fostering an inclusive environment appreciating varied skills, perspectives, and experiences. We are passionate about celebrating diverse voices to support our mission of inspiring creativity and spreading joy within our global community.