Gramian Consultancy is looking for an LLM Agent Trainer with expertise in Traditional Chinese Cantonese. Your work will directly train foundational Large Language Models by creating high-quality proprietary datasets for fine-tuning and benchmarking.
What You'll Do
- Design multi-turn conversations that simulate real interactions between users and AI assistants using applications like calendar, email, maps, and drive.
- Emulate both the user and the assistant, including the assistant's tool calls when corrections are needed.
- Carefully select when and how the assistant uses available tools, ensuring logical flow and proper usage of function calls.
- Craft dialogues that demonstrate natural language, intelligent behavior, and contextual understanding across multiple turns.
- Generate examples showcasing the assistant’s ability to complete feasible tasks, recognize infeasible ones, and maintain engaging general chat.
- Ensure all conversations adhere to defined formatting and quality guidelines using an internal playbook.
- Iterate on conversation examples based on feedback to improve realism, clarity, and training value.
- Collaborate with peers and reviewers to maintain consistency and high standards.
What We're Looking For
- 3+ years of overall professional experience in a technical or analytical field.
- Native, bilingual, or professional proficiency in Traditional Chinese Cantonese.
- Strong grasp of APIs, data formats like JSON, and logical thinking.
- Strong general technical reasoning skills and the ability to model real-world assistant behavior using tool-based APIs.
- Ability to break down complex tasks and simulate realistic dialogues reflecting user expectations and assistant limitations.
- Excellent written communication skills in English, with a focus on clarity, tone, and instructional coherence.
- Creativity and attention to detail in crafting realistic scenarios and responses.
- Ability to follow detailed guidelines and formatting standards with high consistency.
Nice to Have
- Experience working with or around LLMs, virtual assistants, or function-calling frameworks.
Work Mode
This is a global position.
