Applied Scientist, GenAI Catalog Intelligence, PRISM
Core
Build the intelligence layer for a conversational AI agent (Catalog Diagnostic Assistant) that autonomously investigates catalog anomalies across billions of products using agentic architectures and GenAI.
Role type
Founding Applied Scientist (GenAI & Agentic Systems)
Builds
Autonomous diagnostic agent for Amazon's catalog operations
Domain
E-commerce, Generative AI, Agentic Systems, Large-scale Information Retrieval
Deliverable
production ML models
Required skills
PhD or Master's + 4 years in CS/ML, Java/C++/Python, ML model development, agentic reasoning, tool orchestration, RAG systems, model deployment optimization, uncertainty calibration, experimental design
Preferred skills
LLM/VLM deployment on GPUs/Neuron/TPU, explainable AI, statistical analysis, multimodal pretraining/fine-tuning, RLHF, top-tier research publications
Technologies
Python, Java, C++, LLMs, VLMs, RAG, GPUs, Neuron, TPU
Responsibilities
Formulate open research problems in agentic reasoning, design novel agentic architectures, build and optimize RAG systems, advance efficient model deployment strategies, ensure model reliability for autonomous decisions, own research lifecycle from formulation to production, partner with engineers on deployment, shape team research vision, mentor engineers
Seniority
Senior, hands-on IC with research leadership