About this role
Engineers who thrive on building foundational tools and multiplying the output of their peers will find a strong fit in this AI Platform Engineer position at Starburst. The team develops the AI layer for company products, including AIDA, creating conversational agents that provide accurate answers grounded in enterprise data while operating with rapid startup speed inside an established enterprise. As the inaugural dedicated platform engineering hire on the group, the person in this position will construct shared development primitives, robust SDKs, and clean abstractions bridging the Python and JVM environments.
The daily work centers on enabling multiple product teams to ship features faster by owning the complete developer experience for agent development from the ground up. Responsibilities include designing model gateways, establishing API contracts, and creating local debugging tools that ensure high performance and reliability across distributed systems. Success is measured directly by the velocity and productivity of fellow engineers rather than individual feature delivery.
What you will do
- Design and build an agent development SDK that abstracts common patterns such as tool registration, context management, evaluation hooks, and prompt management.
- Construct the language bridge between the Python AI ecosystem and the JVM Starburst platform, ensuring type safety, error propagation, and cross-boundary observability.
- Create comprehensive developer tooling including local development environments, debugging utilities, test harnesses, and CI integration.
- Define and maintain strict API contracts between the AI layer and the core Starburst platform.
- Build shared infrastructure encompassing model gateways, observability pipelines, and experiment frameworks.
- Develop unified, product-agnostic components that can be utilized across multiple internal product lines.
- Own overall platform health beyond feature delivery by prioritizing performance, reliability, and developer ergonomics.
What you need
- Bring at least five years of professional software engineering experience, including a minimum of two years focused on platform, infrastructure, or developer tooling.
- Demonstrate proficiency in both Python and JVM languages such as Java or Kotlin.
- Possess hands-on experience designing APIs and SDKs that are successfully consumed by other engineering teams.
- Demonstrate strong distributed systems fundamentals covering RPC, serialization, error handling, and observability.
- Show practical experience with containerization, CI/CD pipelines, and cloud infrastructure, with a preference for AWS.
- Provide a documented track record of building tools that measurably improve the velocity of other engineers.
Nice to have
- Experience with LLM integration patterns including model APIs, token management, streaming, and tool use
- Experience with cross-language interop such as gRPC, FFI, and language bridges
- Open-source contributions to developer tooling or infrastructure projects
- Familiarity with A/B testing frameworks and experiment infrastructure
- Experience supporting both research and production workloads on the same platform
Benefits
- Competitive pay
- Attractive stock grants
- Flexible paid time off
Skills and tech stack
Questions candidates ask
What are the travel requirements for this role?
This role will require 25% in-person travel for purposes including but not limited to new hire onboarding, team and department offsites, customer engagements, and other company events. Actual travel expectations may vary by role and business needs.
What languages and technologies will I work with?
You will work across the Python and JVM ecosystems, utilizing languages such as Java or Kotlin, alongside distributed systems components, CI/CD pipelines, and cloud infrastructure with a preference for AWS.
What is the primary focus of this team?
The team builds the AI layer for Starburst's products, including AIDA, by designing agents that let users ask questions in natural language and get accurate, grounded answers backed by their actual data.