Senior Applied Scientist, AI Customization & Evaluation

Amazon Web Services (AWS); New York (NY); In loco; USD 184.000 - 249.000; Tempo pieno

Descrizione del lavoro

Amazon Web Services, Inc. in New York seeks a Senior Applied Scientist to guide enterprises in customizing foundation models, from data curation to evaluation and production deployment.

You will apply methods ranging from parameter-efficient fine-tuning to reinforcement learning, and produce reusable recipes and open-source accelerators. The role splits roughly 50/50 between direct customer engagements and building leverage through reference implementations, open-source projects, and thought

Competenze

• PhD or Master’s degree required or equivalent research experience.
• 6+ years of applied research experience in ML.
• 4+ years building ML models for business applications.

Mansioni

• Design the evaluation methodology for model selection and customization decisions.
• Lead customization engagements end to end from data curation to fine-tuning and reinforcement learning.
• Invent new techniques and package them as reusable code and pull requests for open-source tools.
• Turn engagement learnings into reusable assets, reference architectures, and workshops.
• Advise field teams on trade-offs among prompting, retrieval, and customization of models.
• Mentor engineers and Solutions Architects across teams.

Conoscenze

PhD or MS + 6+ yrs

Java, C++, Python

ML model building 4+ yrs

Formazione

PhD or Master's degree

Strumenti

R

scikit-learn

Spark MLlib

MxNet

TensorFlow

NumPy

SciPy

Descrizione del lavoro

Amazon Web Services, Inc. in New York seeks a Senior Applied Scientist to guide enterprises in customizing foundation models, from data curation to evaluation and production deployment.

You will apply methods ranging from parameter-efficient fine-tuning to reinforcement learning, and produce reusable recipes and open-source accelerators. The role splits roughly 50/50 between direct customer engagements and building leverage through reference implementations, open-source projects, and thought