AI Trainer

Job Purpose
We are seeking highly detail-oriented and technically fluent individuals for a contract-based AI
Trainer role focused on Agentic AI Testing. You will work with simulated conversations and
test intelligent assistant behavior across multi-turn interactions using advanced tools like Slack,
Jira, PayPal, and Amazon APIs.

Key Responsibilities
As an AI Trainer, your role involves creating and curating high-quality training data for
agent-based assistants. You will:
● Design realistic, multi-turn conversations where an AI assistant uses tools to solve
complex user goals.
● Simulate both user and assistant roles, writing conversations that reflect natural dialogue
and require tool-based reasoning.
● Use applications backed by RL Gym databases including: Slack, Jira, PayPal, Amazon.
● Ensure all assistant responses align with real-time tool data, API structures, and correct
function invocation.
● Apply error labeling and correction to model failures (e.g., hallucinations, incorrect tool
use).
● Create fallback behaviors, datetime reasoning, and edge-case instructions.
● Adhere to detailed formatting and structural guidelines including tool chaining, system
prompt design, and review checklists.

● Maintain logical coherence and tool output grounding across 10–15 turn dialogues.
● Follow standards for natural language, prompt injection detection, and context
reasoning.

Minimum Qualifications
● Bachelor’s degree in Computer Science, Engineering, or a related field.
● Strong command of English writing and logic.
● Familiarity with APIs, JSON formats, or prompt-based tools.
● Ability to reason through structured data and simulate human-like interactions.

Preferred Skills
● Experience with conversational AI, LLMs, or prompt design.
● Exposure to tools such as JIRA, Slack, Amazon Seller Central, or PayPal APIs.
● Strong attention to detail and ability to follow structured playbooks and feedback loops.

Compensation & Work Environment
● 💰 Rate: $20/hour
● 🌍 Location: Remote
● ⏱ Schedule: Flexible hours; task-based evaluation
● 🧠 Work on cutting-edge AI datasets with clear impact on real-world assistant behavior.