Applied AI Scientist, Small Language Model and AI Training

Postman Postman · Enterprise · San Francisco, CA · AI

Applied AI Scientist role focused on research and development of small, efficient language models, including training methodologies, optimization, compression, fine-tuning, and ensuring AI safety and alignment. Collaborates with research, engineering, and product teams on scalable training pipelines and deployment.

What you'd actually do

  1. Lead research and development of novel training methodologies and architectures for small and efficient language models.
  2. Design, implement, and evaluate model training experiments to improve performance, robustness, and generalization of language models.
  3. Collaborate closely with research scientists and engineers on scalable training pipelines and model deployment strategies.
  4. Develop techniques for model compression, fine-tuning, and domain adaptation to optimize models for real-world applications.
  5. Ensure AI safety, fairness, and alignment principles are integrated into model training processes and evaluated rigorously.

Skills

Required

  • Python
  • PyTorch, TensorFlow, or JAX
  • language model architectures
  • training techniques
  • optimization strategies
  • distributed training
  • data pipeline design
  • scalable AI infrastructure

Nice to have

  • large and small language models in production or research settings
  • reinforcement learning
  • prompt engineering
  • transfer learning techniques
  • developer tools, APIs, or frameworks related to AI model integration and delivery
  • AI alignment, fairness, and ethical AI training methodologies

What the JD emphasized

  • small and efficient language models
  • model training experiments
  • scalable training pipelines
  • model compression, fine-tuning, and domain adaptation
  • AI safety, fairness, and alignment principles
  • distributed training, data pipeline design, and scalable AI infrastructure

Other signals

  • building efficient, high-performance language models
  • advance model training techniques
  • optimize architectures
  • scale AI solutions
  • AI systems that are safe, interpretable, and impactful