Haotian L.

Key Skills

Software

Internal/Proprietary Tooling

Scale AI

Snorkel AI

Top Subject Matter

Enterprise Software – AI Agent Evaluation & Prompt Quality

E-commerce & Supply Chain – Procurement & Sourcing Automation

Technology – LLM Output Assessment & Data Labeling

Top Data Types

Image

Text

Document

Top Task Types

Classification

Text Generation

Question Answering

Text Summarization

Freelancer Overview

I'm an AI Product Manager at Meta with hands-on experience building, evaluating, and refining AI systems at enterprise scale. My work involves defining and assessing AI output quality across dimensions including groundedness, correctness, helpfulness, and safety — the same criteria central to AI training and data labeling work. I designed evaluation frameworks for Meta's internal AI agents, establishing rubrics to measure whether responses were accurate, contextually appropriate, and free from hallucination before deployment to 70,000+ employees. Beyond evaluation, I've built RAG-based AI systems where I directly shaped the knowledge grounding layer — curating source documents, testing retrieval accuracy, and iterating on prompt design to improve response fidelity. I've also led automated case triage work involving intent classification, sentiment labeling, and category assignment at scale. This gives me practical fluency in the exact tasks AI trainers perform: judging response quality, identifying failure modes, writing preference annotations, and providing structured feedback that meaningfully improves model behavior. I bring strong analytical skills, precise written communication, and a genuine passion for making AI systems more reliable and beneficial.

Education

Work History