Engineering Analyst, Trust and Safety, Youtube

Google Google · Big Tech · Hyderabad, Telangana, India

This role focuses on using and improving LLMs for content moderation and abuse prevention on YouTube. Responsibilities include designing and refining prompts for LLMs, analyzing data to understand abuse impact, and driving anti-abuse experiments. The role requires experience in data analysis, project management, and programming languages like Python or SQL, with a preference for machine learning systems experience.

What you'd actually do

  1. Perform fraud and spam investigations using multiple data sources, identify product vulnerabilities and drive anti-abuse experiments to prevent abuse. Work with engineers and interact cross-functionally with stakeholders to improve workflows by process improvements, automation and anti-abuse system creation.
  2. Design and refine prompts for Large Language Models (LLMs) to improve their accuracy in identifying and classifying abusive content and behavior, including prompt engineering, data labeling, and performance analysis.
  3. Apply advanced statistical methods to datasets to understand the impact of abuse to the YouTube ecosystem. Contribute strategy and development of new workflows.
  4. Learn technical concepts and systems and deliver results using them.
  5. Maintain and promote quality by providing regular feedback metrics to the Global team. Manage technological solutions for streamlining quality assurance and produce training solutions.

Skills

Required

  • data analysis
  • identifying trends
  • generating summary statistics
  • drawing insights from quantitative and qualitative data
  • managing projects
  • defining project scope, goals, and deliverables
  • SQL
  • R
  • Python
  • C++

Nice to have

  • machine learning systems
  • collecting, managing and synthesizing datasets and information from disparate sources
  • statistical modeling
  • data mining
  • data analysis
  • metrics analysis
  • experiment design
  • automation
  • communication skills

What the JD emphasized

  • anti-abuse experiments
  • prompt engineering
  • LLMs

Other signals

  • prompt engineering
  • anti-abuse experiments
  • LLM accuracy
  • statistical methods
  • automation