Zaheer H.

Key Skills

Software

Appen

Other

Don't disclose

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Text

Top Task Types

Question Answering

Segmentation

Text Summarization

Translation/Localization

Freelancer Overview

AI Annotation & Data Labeling Expertise – Temu Project Led critical AI training operations within Temu’s chat support framework, focused on enhancing automation accuracy and decision-making efficiency. Supervised large-scale annotation teams across the following key modules: AI Decision Model Reviewed AI-generated response options and selected the most appropriate reply for customer inquiries. Ensured AI systems made contextually accurate decisions by minimizing confusion and training models on real-case response selection. Supervised and enhanced AI-based decision workflows to ensure reliability and reduce miscommunication. Intent Modeling Analyzed entire customer messages to understand the underlying intent behind queries. Tagged messages based on purpose, enabling the AI to better comprehend natural language and customer expectations. Improved AI’s intent recognition by accurately labeling customer inputs across various use cases. Scene Selection In scenarios where customers expressed issues over multiple messages, encapsulated the entire context into a single, meaningful tag. This helped the AI capture the overall sentiment or problem when information was fragmented. Led tagging of multi-turn conversations to ensure end-to-end scene comprehension. Customer Service Annotation Handled escalated tickets where AI failed to resolve customer issues. As the "human next to AI," re-evaluated cases using all evidence and applied corrective labels to train the AI for

Labeling Experience

Temu - Customer Chat AI Annotation

Don't discloseTextQuestion AnsweringText Summarization

Scope of the Project: The project focused on building and refining Temu’s AI-based chat support system. It involved multiple annotation categories including AI decision validation, intent modeling, scene selection, and customer service complaint resolution. Each category had strict accuracy and quality KPIs, with a required accuracy rate of 92%. Agents were expected to meet a minimum daily target of 2,000 labeled cases, ensuring high-quality data to train and improve AI performance. Project Size: The project team consisted of approximately 100+ agents, including QA analysts and team leads, operating under a structured BPO environment(Mindbridge). The large-scale annotation effort supported real-time AI learning for one of the fastest-growing e-commerce platforms globally.

2024 - 2025

Education

Work History