AI Cloud Data Architect

About

Dice is the leading career destination for tech experts at every stage of their careers. Our client, Photon, is seeking the following.

Responsibilities

• Hands-on player-coach role: Owns architecture and directly contributes to AI/ML build (prompt engineering, RAG patterns, agent tuning) must be staffed with a senior who has shipped LLM systems on Bedrock, not a pure-architecture profile.
• Own the end-to-end target architecture for AI use cases on AWS stack (100% AWS), with the data lake on Snowflake.
• Design and hands-on build of solutions on AWS Bedrock the foundation confirmed is already in place for the servicing agent and is the strategic agentic substrate for SLM. Includes prompt engineering, retrieval pipelines (RAG), and foundation model selection from Bedrock's catalog.
• Implement Bedrock Guardrails ( "guardrails is AWS's Bedrock") for content safety, PII handling, and accuracy thresholds non-negotiable in regulated lending.
• Ensure architectural alignment with Sallie Mae's existing MuleSoft investment (full API lifecycle: design, testing, runtime, governance) and the planned MuleSoft Agent Fabric control-plane the SOW pod must integrate, not duplicate.
• Deliver agent layer code in SLM-preferred framework (assumption: LangChain / LangGraph or AWS Strands, depending on Anupam's selection).
• Drive eval-harness design (paired with QE/AI Testing Engineer for execution) covering agent accuracy, hallucination rates, and regression. Specifically flagged evals: "how do you make sure that your agents continue to perform with a high level of accuracy as underlying things change."
• Collaborate with VP, Technology Enablement, on the broader modernization roadmap (50%+ of private student lending applications need modernization).
• Own architecture decision records (ADRs) and engagement with the enterprise architecture function.
• Coordinate the future integration roadmap: CRM integration (Salesforce), omnichannel exposure (web + mobile), and IVR none of which are wired up yet.