Applied AI Demo Engineer

About the role

Anthropic's customer-facing teams are often the first place a customer sees what Claude can do. As a Demo Engineer, you'll build the demos, reference implementations, and reusable tooling that help those teams make new capabilities tangible, relevant, and trustworthy, ideally within days of a product launch rather than weeks after.

This role sits between product engineering, technical storytelling, and GTM execution. You'll own ambiguous, high-leverage problems end to end: prototyping an agentic experience for a new capability, then hardening it into something reliable that other teams can run on their own. You'll work closely with Product, Engineering, Marketing, Operations, and our sales and solutions teams, and you'll bring what you learn from customer-facing work back to the people building the products.

Because a demo is frequently someone's first impression of what our models can and can't do, accuracy and honesty matter as much as polish. Success in this role looks like customer-facing teams demonstrating new capabilities sooner and with more confidence, recurring requests becoming reusable assets rather than one-off projects, and field insights shaping product and launch-readiness decisions.

Responsibilities

• Own demo readiness for new and priority Claude capabilities, including environments, integrations, synthetic data, evaluations, reliability checks, and fallback paths
• Build prototypes, agents, and reference flows that make complex capabilities concrete and tie them to clear customer outcomes
• Turn recurring requests into reusable accelerators such as companion apps, templates, Skills, plugins, and self-service demo packages that scale across customer-facing teams
• Partner with GTM colleagues on high-impact customer moments, unblock critical technical gaps, and spot the patterns worth turning into shared assets
• Build internal agentic systems that automate demo preparation, validation, and asset maintenance, so demos stay current as products change
• Translate patterns from customer-facing work into product feedback, launch-readiness priorities, and systemic improvements with Product and Engineering
• Define and raise the bar for demo quality across accuracy, safety, reliability, usability, and adoption
• Decide when a customer need calls for a bespoke build and when it should become a shared capability

You may be a good fit if you have

• Proficiency in Python or TypeScript, with experience shipping working software end to end
• Hands-on experience building applications or agents on top of LLM APIs, including prompting, tool use, and retrieval
• Experience designing integrations with third-party APIs, data sources, and services
• The ability to take an ambiguous capability or customer need to a working prototype quickly, and then make it reliable
• Clear communication of technical concepts to both technical and non-technical audiences, including sales and solutions teams
• Comfort collaborating across Product, Engineering, Marketing, Operations, and GTM teams
• Experience in a demo engineering, solutions engineering, sales engineering, forward deployed engineering, developer relations, or product engineering role
• A track record of building reusable internal tools, templates, or self-service assets that other teams adopted
• Hands-on experience with Claude Code, the Claude Agent SDK, the Model Context Protocol (MCP), Skills, or plugins
• Experience building evaluation or validation harnesses for AI systems, including reliability checks and graceful fallbacks
• Experience building always-on or scheduled agentic systems that automate internal workflows
• Experience building synthetic or sanitized datasets that make demos realistic without exposing sensitive information
• Front-end or product design skills that help demos feel polished and intuitive
• Experience supporting product launches or running launch-readiness programs
• A history of turning field feedback into concrete product input that influenced a roadmap
• A passion for making powerful technology safe and beneficial, and a commitment to representing model capabilities accurately and honestly