Principal/senior Machine Learning Scientist - Search and Discovery

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

Roblox is seeking experienced machine learning researchers to innovate on LLMs, VLMs, and other large models for user understanding, content understanding, recommendation, and agentic search. The role involves designing and implementing ML solutions for agentic search and discovery systems, large-scale recommendation models, and LLMs/VLMs for visual and natural language understanding. The position requires a practical approach to integrate research into products and experience with transformer-based model training, inference, and product integration.

What you'd actually do

  1. Design and implement the machine learning solutions that power agentic search and discovery systems to drive Roblox content distribution.
  2. Design and implement large scale recommendation models.
  3. Design and implement LLMs and VLMs for visual understanding and natural language understanding.
  4. Balance researching new technologies with a practical approach to integrate research efforts into the Roblox products.
  5. Collaborate with other teams within Roblox to make sure we are building products with a community first approach.

Skills

Required

  • transformer based model training
  • transformer based model inference
  • transformer based model product integration
  • designing and building scalable machine learning solutions

Nice to have

  • LLMs
  • VLMs
  • autoregressive models
  • user understanding
  • content understanding
  • recommendation
  • agentic search
  • visual understanding
  • natural language understanding
  • large scale recommendation models
  • agentic search and discovery systems

What the JD emphasized

  • 7+ (for principal) or 4+ (for senior) years of experience designing and building scalable machine learning solutions.
  • Experienced with transformer based model training, inference, and product integration.
  • Break down product requirements into iterative deliverable stages.

Other signals

  • LLMs
  • VLMs
  • recommendation models
  • agentic search
  • transformer based model training, inference, and product integration