Description:

Amazon is hiring an Applied Scientist I for its Trustworthy Shopping Experience (TSE) Product team in Bengaluru. Operating at the cutting edge of Generative AI, this elite R&D role aims to guarantee that hundreds of millions of products sold on Amazon are safe, authentic, and compliant with global regulations. You will transition manual, human-in-the-loop risk investigation workflows into autonomous, intelligent systems. Acting simultaneously as a researcher and an SDE I-level builder, you will map business goals to hard scientific metrics—inventing, training, and productionizing multi-step Agentic AI reasoning systems, self-supervised representation architectures, and multimodal fusion frameworks that seamlessly parse text, product images, and relational database signals.

Key Technologies:

Generative AI, Large Language Models (LLMs), Agentic Architectures, Natural Language Processing (NLP), Computer Vision, Multimodal Fusion Systems, Supervised Fine-Tuning (SFT), Reinforced Fine-Tuning (RFT), Reinforcement Learning (RL), Python, Java, C++, SQL, Relational Databases (RDBMS), Data Warehouses

Requirements:

• Preferred candidates possess a strong quantitative advanced degree (Master's or PhD) with a specialized focus on Machine Learning, NLP, Computer Vision, or a highly related technical field. A record of peer-reviewed publications at top-tier AI conferences (e.g., NeurIPS, ICML, CVPR, ACL) is a significant advantage.
• Proven software development experience writing secure, stable, testable, and maintainable production code in Python, Java, or C++ at an Amazon SDE I proficiency level.
• Hands-on exposure configuring AI agents capable of multi-step autonomous reasoning, memory retention mechanisms, and policy optimization leveraging Reinforcement Learning (RL) paradigms.
• Baseline technical familiarity productionizing models using Supervised Fine-Tuning (SFT), Reinforced Fine-Tuning (RFT), and zero/few-shot multimodal adaptation.
• Understanding of deep learning techniques used to align and fuse disparate data inputs—specifically combining unstructured text, relational telemetry, tabular documents, and product image arrays into coherent semantic vectors.
• Strong conceptual understanding of foundational computer science data structures, execution complexity algorithms, query creation via SQL, and data retrieval over enterprise RDBMS or multi-tenant Data Warehouses.
• Ability to prototype rapidly, construct secure multimodal data pipelines, run model compression boundaries to minimize cloud deployment costs, and execute deep root-cause failure analysis (RCA) on edge cases.
• Clear verbal and written technical communication skills to publish internal design whitepapers, defend engineering choices during peer code reviews, and guide or mentor incoming science interns.