Member of Technical Staff - Machine Learning (ai Team)

Microsoft Microsoft · Big Tech · Mountain View, CA +4 · Software Engineering

This role focuses on creating and improving LLM models for general-purpose capabilities and products, with a strong emphasis on agentive applications and safety alignment. The responsibilities include developing new training methods, collecting data, evaluating LLMs, building data flywheels, creating tooling for training/evals, writing production code, and developing new user-facing features. The role involves working on core LLM capabilities, fine-tuning, and training classifiers to support Microsoft products and APIs, with a specific focus on advancing towards 'Humanist Superintelligence' that is controllable and safety-aligned.

What you'd actually do

  1. Own and pursue a research agenda to improve model capability and performance for agentive application.
  2. Collaborate closely with the other research and product teams, from pretraining to model hosting to unlock new model capabilities.
  3. Build robust evaluations for tracking modeling improvements.
  4. Design, implement, test, and debug code across our research stack.

Skills

Required

  • Bachelor's Degree in Computer Science or related technical field
  • 2+ years technical engineering experience with coding in languages including, but not limited to, C, C++, C#, Java, JavaScript, or Python

Nice to have

  • Master's degree in Computer Science or related technical field
  • 1+ years technical engineering experience with coding in languages including, but not limited to, C, C++, C#, Java, JavaScript, or Python
  • Doctorate in Computer Science, Machine Learning, Human-Centered AI or related field
  • experience in (e.g., finetuning models with supervision or reinforcement learning, understanding and fixing data quality and curation, working with collaborators on creating new products)

What the JD emphasized

  • agentive application
  • agentive
  • controllable
  • safety-aligned

Other signals

  • LLM models
  • agentive
  • foundation models
  • humanist superintelligence
  • safety-aligned
  • controllable AI