Lead Applied Scientist: Human-AI Collaboration & Evaluation

Amazon Inc.; Austin (TX); Sur place; USD 199 000 - 269 000; Plein temps

Résumé du poste

Amazon AWS Applied AI Solutions (AAIS) seeks a scientist to own the scientific agenda for human-AI collaboration across agentic and multimodal systems. You will turn fuzzy behavioral goals into end-to-end plans, write code weekly, and build large parts of the experimental stack yourself.

Initially hands-on with product and engineering, the role will later join a product team to drive adoption. You will mentor scientists and engineers and represent the work in internal reviews and external venues.

Qualifications

• PhD or Master's in a relevant technical field is required.
• Experience developing agentic AI systems end-to-end.
• Experience designing evaluation for systems without standard benchmarks.
• Proven ability to write substantial production code in Python.
• Experience leading complex scientific work with engineering and product partners.

Responsabilités

• Own the scientific strategy for human-AI collaboration across agentic and multimodal systems.
• Convert desired interactions into falsifiable hypotheses and system requirements.
• Build the evaluation system measuring behavior, trust, adaptation and human contribution.
• Design agent systems using memory, tools, planning and recovery; test with users.
• Design data collection and curation for language, speech and traces.
• Build significant parts of the experimental stack yourself.
• Diagnose failures across data models, orchestration, evaluation and product interaction.
• Partner with design, product, engineering and behavioral research.
• Run experiments with partner product teams and report outcomes.
• Set the standard for reproducible experiments and scientific review.
• Mentor scientists and engineers while staying hands-on.
• Represent the work in internal reviews and external venues.

Connaissances

Agentic AI systems

Multimodal models

Speech systems

Experiment design

Mentoring scientists

Formation

PhD or MS in CS/ML/AI

Outils

Python

Description du poste

Amazon AWS Applied AI Solutions (AAIS) seeks a scientist to own the scientific agenda for human-AI collaboration across agentic and multimodal systems. You will turn fuzzy behavioral goals into end-to-end plans, write code weekly, and build large parts of the experimental stack yourself.

Initially hands-on with product and engineering, the role will later join a product team to drive adoption. You will mentor scientists and engineers and represent the work in internal reviews and external venues.