Principal Machine Learning Engineer, Content Safety

Roblox Roblox · Consumer · San Mateo, CA · Machine Learning

Principal ML Engineer focused on Content Safety at Roblox, defining and leading the technical strategy and execution for large-scale ML systems to proactively detect and mitigate violative UGC content. The role involves end-to-end product development, including data construction, auto-labeling pipelines, and shipping solutions, with a significant portion of time (30-40%) dedicated to backend and integration work. Requires expertise in Computer Vision and/or Vision-Language Models, scalable ML inference services, and robust data pipelines.

What you'd actually do

  1. Define and lead the multi-year technical vision, architectural strategy, and execution for machine learning solutions in Content Safety, ensuring these systems proactively and effectively detect and mitigate violative content at massive scale.
  2. Collaborate with executive-level Product, Data Science, Policy, and Operations leaders to define and prioritize the strategic machine learning roadmap, influencing product strategy and demonstrating the impact of ML on user trust and safety outcomes.
  3. Oversee the adoption and safe deployment of innovative machine learning techniques (e.g., transfer-learning, self-supervised learning, quantization, LoRA, distillation).
  4. You will work cross-functionally to construct datasets from scratch where none exist, build auto-labeling pipelines, and ship solutions to solve novel technical problems.
  5. Expect to spend roughly 30-40% of your time on backend and integration work. You will be responsible for integrating your work into the production stack, leveraging modern AI coding tools (e.g., Cursor) to accelerate velocity and handle infrastructure complexity

Skills

Required

  • 8+ years of experience designing, developing, and operating large-scale, high-impact machine learning systems in a production environment
  • Proven track record of successfully setting the long-term technical direction for an entire ML domain
  • Deep expertise in advanced ML architectures and techniques, including Computer Vision (CV) and/or Vision-Language Models (VLMs)
  • Expertise in architecting scalable, real-time ML inference services and robust data pipelines
  • Demonstrated success in leading and resolving high-stakes, cross-functional conflicts and technical disagreements
  • Exceptional product sense and strategic planning ability

Nice to have

  • transfer-learning
  • self-supervised learning
  • quantization
  • LoRA
  • distillation
  • modern AI coding tools (e.g., Cursor)

What the JD emphasized

  • multi-year technical vision
  • architectural strategy
  • execution
  • massive scale
  • long-term technical direction
  • ambiguous problems
  • scaled production impact
  • real-time ML inference services
  • robust data pipelines
  • high-stakes, cross-functional conflicts
  • technical disagreements

Other signals

  • large-scale ML systems
  • proactive moderation
  • content safety
  • user trust and safety