Senior Applied AI Engineer (Remote)

Job summary

This role involves shaping the development of AI-powered product features by collaborating with software engineers, data engineers, and data scientists. The focus is on leveraging extensive spend data to create valuable product enhancements for customers.

Qualifications

• Proven experience in delivering customer-facing AI features using LLMs in production environments.
• Deep practical knowledge of RAG system design and the complete retrieval pipeline, including embedding models and vector databases.
• Experience in building evaluation frameworks for LLM-based systems to ensure quality.
• Strong Python programming skills, with a focus on production-grade, tested, and maintainable code.
• Familiarity with building APIs and data retrieval pipelines.
• Ability to reason about data quality and architecture independently.
• Experience with agentic system patterns and public cloud providers (AWS, GCP).
• Willingness to work with Kotlin components as needed.

Responsibilities

• Build and ship AI-powered product features, managing outcomes from start to finish.
• Design and implement end-to-end RAG systems for specific use cases.
• Oversee the entire AI product lifecycle, including prompt design and safe production deployment.
• Create evaluation datasets and automated evaluation pipelines to ensure feature quality.
• Implement observability for AI features, including logging and quality monitoring.
• Collaborate with Product and Design teams to define AI features from first principles.
• Partner with the GenAI Platform team to identify infrastructure needs and promote platform tooling adoption.
• Provide technical leadership and support to other engineers through reviews and collaboration.

Skills

• Strong product instincts with a user-centric approach.
• In-depth understanding of LLMOps, data retrieval, and model evaluation in production.
• Ability to communicate AI trade-offs to non-technical stakeholders.

Education

• Relevant experience in AI engineering or a related field is preferred.