Machine Learning Engineer

Adobe Adobe · Enterprise · San Francisco, CA

Machine Learning Engineer role focused on building and shipping AI-powered features for Adobe's Pro Design products. This involves designing, prototyping, and deploying models for predictive, ranking, generative, and agentic systems, including RAG, embeddings, and fine-tuning. The role also requires owning end-to-end evaluation and building MLOps foundations for model reliability.

What you'd actually do

  1. Design, prototype, and ship models powering product features, from predictive and ranking to generative and agentic systems, including RAG, embeddings, fine-tuning, and in-product copilots for creative use cases.
  2. Own evaluation end to end: define what "good" means, build offline and online evaluation harnesses, and detect quality regressions in production.
  3. Build the data and MLOps foundations that keep models reliable: feature pipelines, experiment tracking, versioning, CI/CD, automated retraining, and monitoring.
  4. Partner with Product, Engineering, and Data Science teams to define problems, scope feasibility, and translate research into shipped experiences.

Skills

Required

  • Python
  • SQL
  • MLOps
  • feature engineering
  • supervised/unsupervised models
  • prediction
  • ranking
  • recommendation
  • personalization
  • LLMs
  • generative models
  • RAG
  • embeddings
  • fine-tuning
  • agent design
  • evaluation
  • experimentation

Nice to have

  • Master's or PhD
  • consumer product analytics
  • creative tools
  • SaaS
  • subscription businesses
  • recommendation systems
  • personalization for visual or creative content
  • Databricks
  • MLflow
  • Feature Store
  • model serving

What the JD emphasized

  • 5+ years building, deploying, and operating ML systems in production at scale
  • Working depth in modern GenAI on top of that: LLMs or generative models in production (RAG, embeddings, fine-tuning, or agent design)
  • Rigorous offline and online evaluation and experimentation design and operation

Other signals

  • shipping models powering product features
  • design, prototype, and ship models
  • RAG, embeddings, fine-tuning, and in-product copilots
  • own evaluation end to end
  • build the data and MLOps foundations