LLM & Agent Algorithm Graduate (Search) - 2027 Start (PhD)

TikTok
Singapore
Full timeFunction: Research & Scientific (R&D)Education: phdSkills: ["Engineering implementation","Collaboration","Communication","Learning ability","Passion"]

Build and optimize search-domain LLM foundations and post-training pipelines, including ultra-long/colloquial pre-training and multimodal representations for image/video and e-commerce. Develop the Agent execution framework (Harness) with multi-agent orchestration, tool calling, planning, memory, and training–inference–evaluation loops. Work on long-term memory and self-improvement (RSI), automated reward/verifier systems, and evaluation frameworks. Apply the work to long-horizon task agents, visual search, and on-device intelligence.

Loading

Loading job details...

Preparing the role view and application actions.

FursaFursa
TikTok
TikTok
14 hours ago

LLM & Agent Algorithm Graduate (Search) - 2027 Start (PhD)

✓ Verified Job

Canonical indexed version, validated from employer's careers page.

Source: Company careers pageValidated by: Fursa AI
Last checked: 14 hours agoStatus: Live

Job Summary

Build and optimize search-domain LLM foundations and post-training pipelines, including ultra-long/colloquial pre-training and multimodal representations for image/video and e-commerce. Develop the Agent execution framework (Harness) with multi-agent orchestration, tool calling, planning, memory, and training–inference–evaluation loops. Work on long-term memory and self-improvement (RSI), automated reward/verifier systems, and evaluation frameworks. Apply the work to long-horizon task agents, visual search, and on-device intelligence.
Location: Singapore
Employment Type: Full time
Job Function: Research & Scientific (R&D)

Key Responsibilities

  • •Build and optimize search-domain LLM foundations and post-training pipelines, including long-context and multimodal representation learning.
  • •Contribute to the Agent execution framework (Harness), including tool calling, planning, memory, environment interaction, and multi-agent scheduling algorithms.
  • •Develop agent loop closed-loop engineering for training–inference–evaluation and implement self-improvement mechanisms such as self-evolve/RSI.
  • •Work on long-term memory mechanisms, cross-context knowledge integration, and data synthesis/quality control for vertical-domain data.
  • •Help build reward/verifier systems and automated evaluation frameworks, and support business implementation for long-horizon task agents and multimodal AIGC/visual search/on-device intelligence.

Key Requirements

  • •Be completing or have recently completed a PhD in computer science or a related discipline.
  • •Demonstrate a solid foundation in machine learning/deep learning and familiarity with modern LLM and agent technologies, with top-tier conference publications or awards as a plus.
  • •Have experience with LLM/agent projects including pre-training/fine-tuning/alignment, tool calling/planning/memory, RAG, RLHF/RLAIF, data synthesis, and automated evaluation.
  • •Be familiar with PyTorch and/or TensorFlow for training/deployment, including distributed training and mixed-precision; understand compression and inference acceleration (e.g., quantization, pruning, distillation, TensorRT).
  • •Have familiarity with big data frameworks and applications such as MapReduce and Spark (a plus).
Education:PhD / Doctorate in computer science
Skills:Engineering implementationCollaborationCommunicationLearning abilityPassion
Tech Stack:PyTorchTensorFlowMapReduceSparkLLMAgentsHarnessRAGRLHFRLAIFTensorRTQuantizationPruningDistillationDistributed trainingMixed-precision trainingMulti-agent collaborationOCRTinyLLM

Company Brief

TikTok
Short-form video platform that lets users create, share, and discover entertainment content through algorithmic recommendations. It also offers advertising and creator tools for brands, influencers, and businesses.
Industry: Digital Media
Company Size: Enterprise (1,001+ employees)
Growth: Scaleup
Headquarters: Singapore, Singapore
Founded: 2016
WebsiteLinkedIn