Solution Architect

AI agents are non-deterministic: the same input can produce different outputs across runs. Designing enterprise automation around them takes a different discipline than deterministic RPA; grounded context, dynamic decision logic, and guardrails instead of fixed, scripted paths.

This is a forward-deployed, customer-facing engineering role. You will work side by side with enterprise and public sector customers to design, build, and harden agentic workflows in their live environments; combining the hands-on coding of a software engineer, the platform depth of a UiPath specialist, and the end-to-end perspective of a solution architect. You will own the question every customer asks before putting an AI agent into production: “How do we know we can trust it?”

WHAT YOU'LL DO AT UIPATH

• Own the end-to-end architecture for UiPath automation and agentic programs at customer sites - agents, RPA workflows, integrations, data, and human-in-the-loop steps - and document the trade-offs behind it.
• Design agentic workflows that combine LLM decision-making with tool and function calling, context grounding, and deterministic fallbacks, so behaviour stays predictable where the business needs it to be.
• Architect Document Understanding pipelines - extraction, classification, validation - and the Python services that connect them to customer systems.
• Choose the right GenAI/ML technique per problem; prompting, retrieval and context grounding, or a trained model in AI Center; and set the guardrails, human checkpoints, and escalation paths before go-live.
• Deploy and operationalize AI agents in production customer environments, including APIs, data pipelines, and integration with existing systems.
• Act as a trusted technical advisor: translate requirements into architecture and explain design choices, risks, and trade-offs to audiences from engineers to executives.
• Set the standards, reusable patterns, and design reviews that govern delivery from discovery to hypercare; mentor developers and keep the architecture something customers can operate themselves.
• Feed field findings such as failure patterns, missing capabilities, design gaps; back into the product roadmap and into reusable assets and accelerators.

…