JOB SUMMARY

The AI Solutions Engineer designs, builds and supports production AI systems. Their duties include agentic workflows, retrieval pipelines, tool calling integrations, and the evaluation frameworks that demonstrate correct behavior. The AI Solutions Engineer receives a scoped problem and a defined outcome, determines the technical approach, and delivers a working solution, raising design options and risks early rather than working from detailed implementation instruction. They independently determine technical approaches, proactively identify risks and design options, and provide technical leadership to support Vantage Bank's AI strategy, governance, and responsible adoption of AI technologies.

ESSENTIAL DUTIES

The duties listed below may not include all responsibilities that the person in this role may be asked to perform. Incumbent may be required to perform other related duties as assigned.

• Designs and builds agentic AI workflows. These are multi step, tool calling systems with controlled state transitions, enforced boundaries on what the system can do, defined human review points, and step level logging sufficient to reconstruct any decision after the fact.

• Builds and tunes retrieval augmented generation pipelines over bank documents and governed data, covering chunking strategy, embedding model selection, hybrid and semantic retrieval, re-ranking, and evaluates retrieval quality separately from answer quality.

• Integrates AI systems with bank data, applications and services through tool calling interfaces and Model Context Protocol servers and clients, under least privilege access, with auditable call logs and a read only default for anything that could write.

• Builds and operates the evaluation tooling that proves these systems work, including benchmark question sets with expected answers, hallucination and grounding checks, access boundary tests run from the requesting user's identity, regression suites across prompt and model versions, and model graded scoring calibrated against human labels.

• Builds and maintains evaluation harnesses alongside other engineers doing this work, both for this role's own solutions and for shared harnesses that teams across the bank rely on, keeping them current as prompts, models and requirements change.

• Deploys, monitors, and supports AI services in production, covering prompt and model versioning, approval routing for higher risk use cases, drift and quality monitoring, usage and cost visibility, alerting, and incident response with root cause follow through.

• Produces the documentation that makes an AI solution defensible in a regulated bank, including solution logic, prompt instructions, verified queries, test cases and results, governance assumptions, known limitations, and plain language guidance on safe use for the business teams who will rely on it.

• Builds every solution in accordance with Vantage Bank's AI strategy, policies and guidelines rather than around them, and brings the engineering perspective to governance questions on assigned use cases.

• Stand up lightweight internal applications such as Databricks Apps or Streamlit to put a capability directly in front of a branch, lending or operations team.

• Builds AI enrichment pipelines using SQL AI Functions or equivalent for summarization, classification and structured extraction, and validates the generated fields before anything downstream consumes them.

• Supports governed natural language analytics on Genie Spaces, including semantic definitions, verified SQL examples and Unity Catalog metadata, where an agentic solution depends on them.

• Other duties as assigned.

QUALIFICATIONS

These specifications are general guidelines based on the minimum experience normally considered essential to the satisfactory performance of this position. The requirements listed below are representative of the knowledge, skill and/or ability required to perform the position in a satisfactory manner. Individual abilities and organizational limitations may result in some deviation from these guidelines.

• Bachelor27s degree in Data Science, Computer Science, Information Systems, Statistics, Business Analytics, or a related quantitative field; equivalent applied experience may be considered.

• 4+ years of prior experience building and delivering software or data systems with 2+ years of that experience in hands-on development of AI or LLM based solutions that reached users in a production or pre-production setting.

• Hands-on experience building agentic and tool-calling systems using frameworks such as LangGraph, LangChain, AutoGen, or Mosaic AI, including structured outputs, function calling, API or MCP integrations, end-to-end implementation, and behavioral control design, with the ability to describe a specific implementation end to end and the controls that constrained its behavior

• Demonstrated expertise in RAG systems, including chunking trade-offs, embedding selection, hybrid retrieval, re-ranking, and evaluation frameworks covering benchmark datasets, hallucination and grounding detection, access-boundary and regression testing, judge calibration, and root-cause analysis to distinguish retrieval from generation failures.

• Strong Python and SQL proficiency with end-to-end ownership, including solution design, risk identification, delivery, and the ability to debug, optimize, explain, and defend code regardless of development tools used.

• Practical understanding of Databricks AI/BI capabilities, Genie Spaces, and Unity Catalog business semantics and governed data access patterns, or comparable metadata and governance platforms.

• Working understanding of governed environments, including least privilege access, audit logging, and change control, with the ability to document solution logic, prompt instructions, verified queries, test cases, governance assumptions, limitations, and user adoption guidance in an audit-ready format.

• Clear communication with non-technical colleagues, including the ability to explain what a system does and what it cannot be trusted to do, to someone who will never read the code.

• Must be self-motivated with strong initiative, high level of accountability, and attention to detail.

• Ability to always maintain a high degree of ethical standards and complete confidentiality.

Preferred:

• Azure OpenAI, OpenAI, Anthropic, or comparable model APIs in a production setting, including cost and latency management, and familiarity with AI Gateway or comparable LLM governance capabilities such as approved-model routing, usage monitoring, audit logging, cost attribution, and rate limits.

• MLflow or a comparable experiment and model registry, continuous integration and delivery using Azure DevOps or GitHub, containerization, observability tooling, and agile delivery practices, along with lightweight application frameworks such as SQL AI Functions, Databricks Apps, Streamlit, or Gradio.

• Understanding of banking or financial services data, including customer, deposit, lending, fraud, AML, compliance, or operational reporting domains, and exposure to responsible AI practices in financial services, including acceptable use boundaries, model risk awareness, explainability expectations, and audit-ready documentation.