LLM & Agent Algorithm Expert - TikTok Search

TikTok

Machine learning

Location

Singapore

Employment Type

Regular

Job Code

A54765

Responsibilities

Team Introduction

TT-Search Algorithm & Applied AI is the algorithm team behind the search business built on TikTok (TikTok Search), with the goal of becoming the search engine of choice for users worldwide. Compared with recommendation systems, which passively infer user intent, search delivers content based on users' discovery motivation — making intent expression more precise — and can in turn feed back into the recommendation engine. We are at an inflection point where the search paradigm is shifting from "retrieval-and-ranking" toward "Agents proactively completing tasks." With large language models and Agents as our two driving wheels, we build the search LLM foundation, the Agent execution framework (Harness), multi-agent collaboration, and self-improvement closed loops, supporting the implementation of business scenarios such as multimodal AIGC, visual search, on-device intelligence, and long-horizon task Agents — spanning POI search, Wish search, automated evaluation, infrastructure, and more.

Job Responsibilities

1. Search LLM Foundation: Lead the R&D and iteration of the search-domain LLM foundation, integrating search knowledge for rapid implementation; own the pre-training / post-training pipeline for search LLMs (ultra-long-text / colloquial-text pre-training, image-text / video multimodal representation, e-commerce product multimodal representation learning) as well as inference optimization (long context, model efficiency, quantization / pruning / distillation / inference acceleration).

2. Agent Foundation & Harness: Own the 0→1 / 1→N build-out of the search Agent execution framework (Harness) — unified orchestration of tool calling, planning, memory, and environment interaction; design multi-agent cluster scheduling and collaboration algorithms (task allocation, dynamic scheduling, communication / alignment / conflict resolution); build the Agent Foundation platform to empower the team's engineers / algorithm researchers / PMs with full-stack, lightweight development and rapid launch (multi-agent / single-agent + skill-based / memory), supporting the core search intelligent assistant and vertical-search scenario modules.

3. Loop Engineering, Long-term Memory & Self-Improvement (RSI): Lead the Agent Loop and the "training–inference–evaluation" closed loop; build long-term memory mechanisms for LLMs, cross-context knowledge integration, causal reasoning, and autonomous concept induction capabilities; implement self-evolve / self-improvement and recursive self-improvement (RSI) mechanisms (automated hyperparameter tuning, training pipeline automation, AI-assisted algorithm design, automated model iteration closed loop).

4. Data Synthesis & Evaluation Systems: Build high-quality vertical-domain data synthesis and quality control (distribution alignment, synthetic data evaluation / filtering / refinement); own the online Reward/Verifier system and superhuman-capability benchmark evaluation (annotation-free automated evaluation, long-cycle complex tasks, cross-domain innovation capability evaluation, multi-agent collaboration evaluation standards).

5. Business Implementation & Scaling

Long-horizon task Agents (persistent intent)
Multimodal AIGC creation: leveraging SOTA models to provide image/video generation capabilities and amplify scale effects; powering the Feed's "ask-after-viewing / create-after-viewing" experiences to strengthen users' proactive mindset;
Visual search & on-device intelligence: object detection, OCR, TinyLLM;
Search content ecosystem / creator ecosystem, etc.

6. Frontier Exploration (plus): Pre-research into next-generation non-Transformer architectures, AGI safety and alignment (goal-consistency constraints for large-scale multi-agent systems, interpretability / auditability / safety mechanisms), and other directions, and driving their integration with business scenarios.

7. Technical Accumulation & Output: Abstract algorithm libraries and interfaces to improve reuse and R&D efficiency; regularly share SOTA models to empower the search team and company-level BUs; mentor and support the growth of team members.

Qualifications

Minimum Qualifications

1. Bachelor's degree or above in Computer Science or a related field, with AI algorithm R&D experience

2. In-depth research and hands-on implementation experience in any of the following directions — LLM / Agent / multimodal / search — with the ability to independently own and lead a technical direction.

3. Mastery of at least one of the LLM / Agent technology stacks:

Full-pipeline LLM: pre-training / SFT / RLHF / alignment / inference optimization / data synthesis;
Full-stack Agent: Harness, multi-agent scheduling, memory mechanisms, Agent Loop, self-improvement (RSI), Reward/Verifier.

4. Strong engineering skills, familiar with distributed training, mixed-precision training, inference acceleration (quantization / pruning / distillation / TensorRT), and large-scale data processing (MapReduce / Spark, etc.).

5. Familiarity with any of the following directions, with representative achievements:

CV & Multimodal: image / video retrieval, classification and recognition, segmentation, object detection, OCR, multimodal large models, self-/unsupervised learning;

NLP: pre-training, NLU, multilingual / cross-lingual learning, NLG, transfer / semi-supervised learning.

6. Familiarity with any of the following directions, with representative achievements:

CV & Multimodal: image / video retrieval, classification and recognition, segmentation, object detection, OCR, multimodal large models, self-/unsupervised learning; NLP: pre-training, NLU, multilingual / cross-lingual learning, NLG, transfer / semi-supervised learning.

Preferred Qualifications

1. Candidates with top-tier conference publications (NeurIPS / ICML / ICLR / CVPR / ICCV / ECCV / ACL / EMNLP / AAAI, etc.), well-known open-source projects, competition awards (Kaggle / COCO / ImageNet / GLUE / CLUE, etc.), or large-scale business implementation experience are preferred.

2. Excellent technical leadership, cross-team collaboration, and communication skills; sound judgment on and genuine passion for the long-term value of search and Agents.

Job Information

About TikTok

TikTok is the leading destination for short-form mobile video. At TikTok, our mission is to inspire creativity and bring joy. TikTok's global headquarters are in Los Angeles and Singapore, and we also have offices in New York City, London, Dublin, Paris, Berlin, Dubai, Jakarta, Seoul, and Tokyo.

Why Join Us

Inspiring creativity is at the core of TikTok's mission. Our innovative product is built to help people authentically express themselves, discover and connect – and our global, diverse teams make that possible. Together, we create value for our communities, inspire creativity and bring joy - a mission we work towards every day.

We strive to do great things with great people. We lead with curiosity, humility, and a desire to make impact in a rapidly growing tech company. Every challenge is an opportunity to learn and innovate as one team. We're resilient and embrace challenges as they come. By constantly iterating and fostering an "Always Day 1" mindset, we achieve meaningful breakthroughs for ourselves, our company, and our users. When we create and grow together, the possibilities are limitless. Join us.

Diversity & Inclusion

TikTok is committed to creating an inclusive space where employees are valued for their skills, experiences, and unique perspectives. Our platform connects people from across the globe and so does our workplace. At TikTok, our mission is to inspire creativity and bring joy. To achieve that goal, we are committed to celebrating our diverse voices and to creating an environment that reflects the many communities we reach. We are passionate about this and hope you are too.