Sr. Associate - Prompt Engineer, Workpaper Review, Scoring and Reporting Capability (TBD TD Bank)

INSPYR Solutions • Davenport, IA

4d • Remote

About The Position

We are seeking an experienced Prompt Engineer to support an AI-enabled transformation initiative within Financial Services. This is a hands-on build role focused on developing LLM capabilities for document review, automated scoring, theme detection, and report generation, with a strong emphasis on citation grounding and hallucination prevention. The ideal candidate has built production LLM applications, not just prototypes, and has deep experience with RAG, prompt architecture, structured outputs, and evidence-grounded AI.

Requirements

• 3+ years of hands-on experience building LLM applications used in production.
• Strong prompt engineering experience, including task decomposition, chaining, structured outputs, and evaluation-driven iteration.
• Production experience with RAG across large or complex document sets.
• Demonstrated expertise in grounding, citations, hallucination control, and abstention behavior.
• Strong Python skills.
• Experience with at least one major model platform/API, including OpenAI, Azure OpenAI, Anthropic, or AWS Bedrock.
• Experience optimizing LLM solutions for cost, latency, and production performance.
• Strong communication and stakeholder management skills.

Nice To Haves

• Experience within Financial Services or another highly regulated, document-intensive environment.
• Familiarity with Internal Audit, risk, compliance, or quality assurance processes.
• Experience with agentic AI patterns, tool use, and AI governance.
• Experience designing AI outputs for non-technical expert reviewers.

Responsibilities

• Design prompt architecture and task decomposition for AI-assisted document review, scoring, and reporting.
• Build and optimize RAG workflows across large, complex document sets, including chunking and retrieval strategies.
• Implement citation grounding so AI-generated conclusions are traceable to source documents.
• Develop guardrails to prevent unsupported or fabricated conclusions and enable appropriate abstention when evidence is insufficient.
• Diagnose gaps between AI outputs and human expert conclusions and continuously improve performance.
• Develop structured output and report-generation capabilities.
• Optimize model selection, cost, latency, and performance for production use.
• Document solution limitations, intended use, and operating controls for regulated environments.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Education Level: No Education Listed