Jie L.

Key Skills

Software

Appen

Top Subject Matter

Technology & Engineering — Industrial Automation, Semiconductor Testing & Smart Inspection

Artificial Intelligence — LLM Evaluation, AI Agent Development & Data Annotation

Computer Vision — Image Processing & Industrial Inspection Data Verification

Top Data Types

Text

Top Task Types

Text Generation

Freelancer Overview

I have practical experience in testing and annotating a mainstream Chinese large language model. In this project, I conducted multi-turn dialogues with the model based on provided functional guidelines, specific personas, and background constraints. My focus was on design-testing queries to evaluate whether the model deviated from its designated role or violated platform compliance policies, followed by documenting and reporting these edge cases. Although my overall experience in this area is still growing, all of my submitted evaluations and annotations were successfully accepted by the platform. I deeply understand how to interact with LLMs, possess strong logical reasoning skills, and pay close attention to detail. I am confident in my ability to handle complex evaluation tasks and help improve model performance.

Labeling Experience

Complex command annotation

TextText Generation

Participated in a high-profile Large Language Model (LLM) evaluation project on the Appen platform, focusing on validating model behavior safety and persona consistency under complex constraints. Key Tasks Handled: Thoroughly analyzed multiple sets of complex prompts (including specific personas, personality traits, designated functions, and background constraints) to understand the model's operational boundaries. Independently designed and executed 3 targeted, adversarial testing queries for each prompt scenario. Conducted multi-turn dialogues to evaluate whether the model deviated from its designated persona or violated platform safety policies. Accurately identified and documented anomalous dialogue samples, performing compliance tagging and feedback reporting as required. Project Scale & Quality Achievements: Completed comprehensive adversarial dialogue testing across over 10 distinct complex prompt scenarios, designing and submitting dozens of high-quality evaluation queries. Strictly followed the project guidelines throughout the execution; all submitted evaluation and annotation data successfully passed review and achieved a 100% acceptance rate.

2025 - 2026

Education

Work History