Software Development Engineer, Amazon Search Navigation AI

The Amazon Search Navigation AI team helps customers go from a broad search intent to the few products that fit their needs, so they can make a confident purchase decision. We generate highly relevant, context-aware, and personalized search refinements. We also decide the order they appear in across Amazon"s Search Navigation experiences, guiding customers to the most relevant results. The team owns Search Navigation"s P0 metrics and defect guardrails, which keep performance and reliability high for hundreds of millions of customers.

We are reinventing search refinement with generative AI. Our recently launched Smart Refinement experience uses large language models (LLMs) served online to turn product specifications into natural-language suggestions that match how customers think, such as "for colder weather" or "lightweight for travel." Each suggestion is personalized to the customer"s context. Next, we are building multi-turn, conversational refinement, where each interaction builds on the last and helps customers narrow their intent in fewer steps.

We build and run the engineering systems behind this work, including the Navigation Experience Service (NES). NES connects to Amazon"s core search systems, serves more than 3 billion queries worldwide every day, and runs real-time LLM inference at global scale under strict latency and cost limits. Our platform supports launches of new navigation content and UIs, and it makes fast offline and online experimentation possible.

We are looking for a Software Development Engineer who is a pragmatic problem solver, comfortable with ambiguity, and curious about applied science. You will build and improve the Search Navigation platform, including the systems that serve LLMs in real time, and work closely with Applied Scientists to take new models from research into production.

Key job responsibilities

1. Design, build, and operate large-scale, distributed, low-latency services that power Amazon Search Navigation experiences worldwide

2. Build and optimize online LLM serving for Smart Refinement and multi-turn conversational refinement, with a focus on latency, reliability, and cost

3. Work with Applied Scientists, Product Managers, and Program Managers to define requirements and integrate models into production systems

4. Build scalable data pipelines that process large volumes of search and behavioral data for feature generation, model training, and evaluation

5. Design, run, and analyze online A/B experiments to measure the impact of new models and features

6. Contribute to operational excellence for critical services, including monitoring, on-call, and P0 metrics and defect guardrails

7. Write high-quality, well-tested code and take part in design and code reviews