Senior Software Development Engineer
Posted Updated
Description
Build the agentic system that automatically improves campaigns for millions of Amazon advertisers. The Sponsored Products and Brands (SPB) auto-optimization framework uses LLM-based agents, together with auction simulation, ML models and optimization algorithms, to decide what to change, apply the change within limits the advertiser sets, measure the results and learn from them. You'll own the architecture and production systems behind this loop, and because the system spends advertiser money automatically, you'll also set its safety standards.
You'll lead the design and delivery of the core services: the agent execution engine and tool orchestration, the action layer that writes changes to live campaigns, evaluation and experimentation pipelines, model serving and tuning infrastructure, and monitoring. You'll work with applied scientists to turn prototype agent capabilities into production systems that run reliably across millions of campaigns. You'll also partner with science, product and operations to improve the agent architecture, success metrics and tuning loops. You'll work backwards from advertisers: every infrastructure decision should make optimization more effective, safer, or easier for advertisers to understand and trust.
What Makes This Role Stand Out
- Agents that act, not just advise: the system makes real changes to live campaigns, so it demands real production engineering
- 1.6MM advertisers: small gains in decision quality have outsized impact on advertiser success and Amazon's retail ecosystem
- Current agent stack: tool orchestration (including MCP), reasoning frameworks, RL fine-tuning and preference optimization
- Safety by design: spend limits, rollback, audit trails and automatic safeguards are core to the architecture
Key job responsibilities
- Own the auto-optimization platform's architecture and drive its engineering roadmap with engineering, science and product leaders
- Build the agent execution engine, covering orchestration, tool invocation, state management and scheduling across millions of campaigns
- Build the action layer that applies agent changes to live campaigns, with spend limits, advertiser controls, conflict handling, rollback and a full audit trail
- Set the safety standards for automated optimization, including observability, drift detection and automatic safeguards that pause optimization when outcomes get worse
- Build offline and online evaluation (simulation, A/B tests, holdouts) to measure decision quality, safety and incremental advertiser impact
- Build a reusable registry of tools and data (auction simulation, ML models, optimizers, campaign data), and integrate tuning pipelines (RL fine-tuning, preference optimization) with interfaces that let scientists version, roll out and roll back models
- Prototype new system designs and agent patterns, use data to decide which go to production, and meet latency, availability and cost targets, including LLM inference cost per advertiser
- Mentor engineers, lead design and code reviews, help with hiring, and own operations (on-call, COEs)
About the team
The Sponsored Products and Brands team at Amazon Ads is re-imagining the advertising landscape through the latest generative AI technologies, revolutionizing how millions of customers discover products and engage with brands across Amazon.com and beyond. We are at the forefront of re-inventing advertising experiences, bridging human creativity with artificial intelligence to transform every aspect of the advertising lifecycle from ad creation and optimization to performance analysis and customer insights. We are a passionate group of innovators dedicated to developing responsible and intelligent AI technologies that balance the needs of advertisers, enhance the shopping experience, and strengthen the marketplace. If you're energized by solving complex challenges and pushing the boundaries of what's possible with AI, join us in shaping the future of advertising.
This team within SPB guides and supports 1.6MM advertisers in creating and managing their ad campaigns. At this scale, the variety of advertiser goals, campaign types and market dynamics is a huge technical challenge and also a big opportunity: even small improvements in guidance systems can have outsized impact on advertiser success and Amazon's retail ecosystem. Our vision is an agentic auto-optimization system that understands each advertiser and keeps improving their campaigns. It combines LLMs with auction simulation, ML models and optimization algorithms. Advertisers stay in control: they choose the instruction to execute and its limits, see what changed and why, and can step in at any time. We work with Ad Console, Sales and Marketing to find where optimization adds the most value, from campaign structure down to individual keyword bids. Our work builds on current agent architectures, tool integration (including MCP), reasoning frameworks, and model customization through tuning and preference optimization.
You'll lead the design and delivery of the core services: the agent execution engine and tool orchestration, the action layer that writes changes to live campaigns, evaluation and experimentation pipelines, model serving and tuning infrastructure, and monitoring. You'll work with applied scientists to turn prototype agent capabilities into production systems that run reliably across millions of campaigns. You'll also partner with science, product and operations to improve the agent architecture, success metrics and tuning loops. You'll work backwards from advertisers: every infrastructure decision should make optimization more effective, safer, or easier for advertisers to understand and trust.
What Makes This Role Stand Out
- Agents that act, not just advise: the system makes real changes to live campaigns, so it demands real production engineering
- 1.6MM advertisers: small gains in decision quality have outsized impact on advertiser success and Amazon's retail ecosystem
- Current agent stack: tool orchestration (including MCP), reasoning frameworks, RL fine-tuning and preference optimization
- Safety by design: spend limits, rollback, audit trails and automatic safeguards are core to the architecture
Key job responsibilities
- Own the auto-optimization platform's architecture and drive its engineering roadmap with engineering, science and product leaders
- Build the agent execution engine, covering orchestration, tool invocation, state management and scheduling across millions of campaigns
- Build the action layer that applies agent changes to live campaigns, with spend limits, advertiser controls, conflict handling, rollback and a full audit trail
- Set the safety standards for automated optimization, including observability, drift detection and automatic safeguards that pause optimization when outcomes get worse
- Build offline and online evaluation (simulation, A/B tests, holdouts) to measure decision quality, safety and incremental advertiser impact
- Build a reusable registry of tools and data (auction simulation, ML models, optimizers, campaign data), and integrate tuning pipelines (RL fine-tuning, preference optimization) with interfaces that let scientists version, roll out and roll back models
- Prototype new system designs and agent patterns, use data to decide which go to production, and meet latency, availability and cost targets, including LLM inference cost per advertiser
- Mentor engineers, lead design and code reviews, help with hiring, and own operations (on-call, COEs)
About the team
The Sponsored Products and Brands team at Amazon Ads is re-imagining the advertising landscape through the latest generative AI technologies, revolutionizing how millions of customers discover products and engage with brands across Amazon.com and beyond. We are at the forefront of re-inventing advertising experiences, bridging human creativity with artificial intelligence to transform every aspect of the advertising lifecycle from ad creation and optimization to performance analysis and customer insights. We are a passionate group of innovators dedicated to developing responsible and intelligent AI technologies that balance the needs of advertisers, enhance the shopping experience, and strengthen the marketplace. If you're energized by solving complex challenges and pushing the boundaries of what's possible with AI, join us in shaping the future of advertising.
This team within SPB guides and supports 1.6MM advertisers in creating and managing their ad campaigns. At this scale, the variety of advertiser goals, campaign types and market dynamics is a huge technical challenge and also a big opportunity: even small improvements in guidance systems can have outsized impact on advertiser success and Amazon's retail ecosystem. Our vision is an agentic auto-optimization system that understands each advertiser and keeps improving their campaigns. It combines LLMs with auction simulation, ML models and optimization algorithms. Advertisers stay in control: they choose the instruction to execute and its limits, see what changed and why, and can step in at any time. We work with Ad Console, Sales and Marketing to find where optimization adds the most value, from campaign structure down to individual keyword bids. Our work builds on current agent architectures, tool integration (including MCP), reasoning frameworks, and model customization through tuning and preference optimization.