Senior Principal AI Systems Architect

Fivetran; Michigan; Vor Ort; USD 150.000 - 190.000; Vollzeit

Zusammenfassung

Fivetran seeks an engineer-architect to make the analyst AI experience real, dependable, and extensible. This role owns the architectur e behind the analyst AI system, including metadata contracts, tool interfaces, context assembly, prompt and skill orchestration, evaluation loops, and the operational patterns that make agents useful in real workflows.

The candidate will define the architecture across prompts, tools, memory and context strategies, evaluation, and runtime safety; build

Qualifikationen

• Strong understanding of LLM prompting, tool use, retrieval, structured outputs, and evaluation methods.
• Deep experience building AI-enabled product systems, not just experiments.
• Comfort working in ambiguous environments where patterns are still being invented.

Aufgaben

• Own architecture behind the analyst AI stack including prompts, tools, metadata, memory and context strategies.
• Define architecture across prompts, tools, metadata, memory and context strategies, evaluation, and runtime safety.
• Build abstractions and interfaces for adding new agent capabilities without brittle prompt sprawl.
• Design retrieval and context strategies with cost, latency and quality tradeoffs.
• Establish evaluation loops, benchmarks, and review processes to measure and improve model behavior.
• Partner with product, engineering, and domain experts to translate workflows into robust AI capabilities.
• Drive quality standards around correctness, trust, transparency, and failure handling in analyst-facing AI experiences.

Kenntnisse

LLM prompting

Software architecture

Tool orchestration

Context strategies

Evaluation methods

Tools

APIs

Contracts

Retrieval systems

Jobbeschreibung

• Fivetran needs an engineer-architect who can make the AI analyst experience real, dependable, and extensible. This role owns the architecture behind the analyst AI system, including metadata contracts, tool interfaces, context assembly, prompt and skill orchestration, evaluation loops, and the operational patterns that make agents useful in real workflows instead of isolated demos
• This is a systems role as much as an AI role. The challenge is not just getting an LLM to do something impressive once. The challenge is building a product-quality analyst system that can reason over schema and business context, use tools correctly, recover from ambiguity, and improve through disciplined evaluation and iteration
• Define the architecture for the analyst AI stack across prompts, tools, metadata, memory and context strategies, evaluation, and runtime safety
• Build the abstractions and interfaces that allow product teams to add new agent capabilities without creating brittle prompt sprawl
• Design retrieval and context strategies that give the model the right information at the right time with strong cost, latency, and quality tradeoffs
• Establish evaluation loops, benchmark suites, and review processes so model behavior can be measured, debugged, and improved systematically
• Partner with product, engineering, and analyst-domain experts to translate real analytical workflows into robust AI-assisted product capabilities
• Drive quality standards around correctness, trust, transparency, and failure handling in analyst-facing AI experiences

Clear judgment about where agent systems should be flexible and where they should be constrainedComfort working in ambiguous environments where patterns are still being inventedPractical understanding of LLM prompting, tool use, retrieval, structured outputs, and evaluation methodsStrong software architecture skills, especially around APIs, contracts, orchestration, and system boundariesAbility to work fluently across backend engineering, product constraints, and user workflow designDeep experience building AI-enabled product systems, not just running experiments