Applied Scientist, Privacy‑Preserving Security ML

Amazon; Seattle (WA); Vor Ort; USD 143.000 - 193.000; Vollzeit

Benefits dieser Stelle

Health insurance

RSUs

401(k) matching

Parental leave

Zusammenfassung

Amazon Leo is recruiting an Applied Scientist to join the founding R&D team for Leo Infrastructure and IP Security. You will build behavioral and statistical models to baseline normal activity across security telemetry from factories, ground stations, and launch facilities.

You will design privacy-preserving data representations, define robust evaluation methods, and ship components to production with rigorous documentation for both scientific and security audiences.

Qualifikationen

• 3+ years building models for business applications.
• PhD or Master’s with 4+ years in CS/CE/ML.
• Proficiency in Python; Rust or C++ a plus; experience with DL frameworks.

Aufgaben

• Build behavioral and statistical models for heterogeneous security telemetry.
• Design privacy-preserving representations of sensitive data.
• Define methodology for difficult, loosely defined problems across domains.
• Develop metrics and evaluation frameworks for models before production.
• Contribute to neurosymbolic reasoning platform and ship production components.
• Document work and communicate results to scientific and security audiences.

Kenntnisse

Statistical modeling

Time-series analysis

Anomaly detection

Behavioral baselining

Python

Rust/C++

PyTorch

TensorFlow

Ausbildung

PhD in CS/CE/ML

Master's degree in CS/CE/ML

Tools

PyTorch

TensorFlow

Rust

C++

Jobbeschreibung

Amazon Leo is recruiting an Applied Scientist to join the founding R&D team for Leo Infrastructure and IP Security. You will build behavioral and statistical models to baseline normal activity across security telemetry from factories, ground stations, and launch facilities.

You will design privacy-preserving data representations, define robust evaluation methods, and ship components to production with rigorous documentation for both scientific and security audiences.