AI Trainer: Multi-Turn Dialogues & Tool Calls

Hire Feed; Pakistan; On-site; PKR 3,715,000 - 7,739,000; Full time

Job summary

Hire Feed is seeking an LLM Trainer - Agent Function call to enhance large language models by generating high-quality proprietary data for fine-tuning and evaluation. You will design multi-turn conversations where AI assistants use function-calling tools to accomplish user goals.

The role emphasizes simulating both user and assistant roles, selecting when tools are used, and iterating on dialogues to improve realism, accuracy, and training value for a global leader in Technology and Information

Qualifications

• Experience with designing multi-turn dialogues and simulating natural language interactions.
• Ability to emulate AI assistants, including tool calls and context handling.
• Familiarity with formatting guidelines and quality standards for high-quality training data.
• Strong attention to detail to craft realistic, coherent interactions.

Responsibilities

• Design multi-turn conversations simulating real interactions between users and AI assistants using tools.
• Emulate both user and assistant roles, including assistant tool calls when corrections are required.
• Select when and how assistants use available tools to ensure logical flow and proper function call usage.
• Craft dialogues demonstrating natural language, intelligent behavior, and contextual understanding across turns.
• Iterate on conversation examples based on feedback to improve realism and training value.

Skills

Multi-turn dialogue design

AI assistant emulation

Training data quality standards

Attention to detail

Job description

Hire Feed is seeking an LLM Trainer - Agent Function call to enhance large language models by generating high-quality proprietary data for fine-tuning and evaluation. You will design multi-turn conversations where AI assistants use function-calling tools to accomplish user goals.

The role emphasizes simulating both user and assistant roles, selecting when tools are used, and iterating on dialogues to improve realism, accuracy, and training value for a global leader in Technology and Information