Machine Learning Engineer (AI Search ), TikTok Search

关于这个机会

Our Search Team focuses on LLM application algorithms for TikTok Search, with core scenarios including AI Search and TikTok AI Bot. We build search question-answer cards via LLM-powered generation and develop TikTok's official AI assistant, leveraging LLM ChatBot, Agent, and RAG technologies to deliver accurate, real-time responses. By closely integrating technical innovation with business scenarios, we bridge users with information and services, driving TikTok's business growth. Responsibilities: - Build AI Search Agents with planning, memory, and tool-use capabilities. Apply reinforcement learning, knowledge representation, and automated decision-making to solve complex problems end-to-end. - Enhance LLM-based query understanding: process multi-turn conversations, conduct semantic analysis on queries and documents, and ensure accurate, objective search results via LLM reasoning. - Optimize RAG and LLM pipelines: improve retrieval and ranking algorithms, train and optimize LLMs for answer generation (e.g., fine-tuning, inference optimization), and enhance the relevance and structured presentation of search results. - Translate research (e.g., Agent/LLM innovations) into production, collaborating with cross-functional teams to drive product upgrades.

任职要求

Minimum Qualifications - Proficient in C++/Python; strong command of algorithms and data structures (e.g., retrieval, graph algorithms). - Basic understanding of LLM fundamentals (e.g., prompt engineering, fine-tuning workflows). - Strong problem-solving skills with a track record of delivering technical projects. Preferred Qualifications - Hands-on experience with LLM/Agent technologies: RAG optimization, RLHF/SFT for LLMs, or Agent framework development (e.g., planning/memory modules). - Excellent communication and collaboration skills, with a track record of exploring new technologies and driving technical advancements with teams.

内推申请说明

本站为独立内推协助平台。申请会交由管理员核实岗位与内推渠道，不等于已在公司官网投递；薪资、岗位状态和实际招聘流程以官方信息为准。