Remote AI Research Accelerator: LLMs & AI Agents

Binance; Singapore; Remote; SGD 17,000 - 26,000; Full time

Job summary

Binance is seeking a driven student or recent graduate to contribute to training, fine-tuning, and evaluating LLMs within a global blockchain ecosystem. You will assist in building AI agents, prompt engineering, memory modules, planning strategies, and integrating external tools.

You will design data annotation pipelines, define schemas, and establish labeling guidelines with quality controls, collaborating closely with research and engineering teams to improve model performance and robustness.

Qualifications

• Pursuing or completed degree in CS/AI/EE; PhD is a bonus.
• Solid ML and DL fundamentals.
• Experience with prompts, LoRA/QLoRA, and model evaluation is a plus.
• English fluency required; other languages a plus.

Responsibilities

• Assist in training, fine-tuning, and evaluating LLMs using datasets.
• Support AI agents development: prompts, memory, planning, tool integration.
• Design and manage data annotation pipelines with guidelines and QC.
• Collaborate with research and engineering to improve models.
• Conduct experiments, data analysis, and document methods and findings.
• Explore new tools and practices to enhance LLM systems.

Skills

Machine learning fundamentals

Deep learning fundamentals

Analytical thinking

English fluency

Education

Degree in Computer Science, AI, or Electrical Engineering

PhD is Bonus

Tools

Transformer models and LLMs

LoRA/QLoRA familiarity

Data annotation tools knowledge

Job description

Binance is seeking a driven student or recent graduate to contribute to training, fine-tuning, and evaluating LLMs within a global blockchain ecosystem. You will assist in building AI agents, prompt engineering, memory modules, planning strategies, and integrating external tools.

You will design data annotation pipelines, define schemas, and establish labeling guidelines with quality controls, collaborating closely with research and engineering teams to improve model performance and robustness.