APPLIED AI SPECIALIST

Frank L Blum Construction Co•Raleigh, NC

3d•Hybrid

About The Position

The Applied AI Specialist uses artificial intelligence to solve practical business problems. The work is applied rather than theoretical: building tooling that assembles reporting from the company's corporate operating systems so that business analytics improve, and developing tools that make each department's day-to-day work move faster and with better information. This is an individual contributor position for someone who can take a defined problem and deliver a working, documented, production-ready solution. The Director of Business Intelligence sets departmental technical standards and carries governance and vendor decisions. Within those standards the person in this seat owns the design of their assigned solution areas.

Requirements

Bachelor's degree in computer science, software engineering, data science, information systems or a related field, or an equivalent combination of education and experience.
Three or more years of professional software, data or automation engineering experience.
Production experience building against large language model services, including prompt design and tool or function calling.
Demonstrated experience building retrieval over a document corpus, including chunking, embedding and relevance evaluation.
Demonstrated proficiency in a general-purpose programming language used for integration work, including work with web interfaces, authentication and data transformation.
Experience connecting third-party software platforms to one another through their published interfaces.
Working proficiency writing queries against a relational database, and a working understanding of data warehouse concepts.
Working practice with source control, automated testing and deployment in a team environment.
Demonstrated practice building evaluation sets and regression tests for language-model applications, and reporting accuracy and cost over time.
Experience operating a language-model application in production, including monitoring, failure alerting and incident response.
Ability to judge whether a problem should be solved with artificial intelligence, with conventional automation, or not at all, and to say so plainly.
Ability to explain technical work to a non-technical audience.

Nice To Haves

Experience with vector databases, embedding model selection or document processing at scale.
Exposure to data warehouse design or analytics engineering.
Experience evaluating artificial intelligence platforms or vendors and writing up the findings.
Experience with agent frameworks and open agent protocols (Model Context Protocol or equivalent), including packaging reusable agent skills for non-technical users.
Experience extracting structured data from engineering drawings, specifications or contracts using vision-capable models.
Familiarity with construction or other project-based operating data: job cost, schedules, submittals, RFIs, change orders, subcontractor bids.
Experience with Microsoft 365, Power Platform or SQL Server as an integration surface.

Responsibilities

Build tooling that assembles reporting from the company's corporate operating systems and provides for better business analytics.
Develop artificial intelligence tools that provide for more efficient workflows within each department's function.
Design, build, test and deploy agents and assistants that automate research, document review and report preparation.
Prototype quickly, measure each result against the process it replaces, and either harden it for production or retire it.
Extract structured data from construction documents (drawings, specifications, contracts, submittments, schedules and meeting transcripts) using vision-capable and language models, so that downstream scheduling, estimating, contract and risk tools answer from the project record rather than from general knowledge.
Write and maintain the integrations required to connect the company's platforms to one another and to artificial intelligence services.
Design and maintain retrieval pipelines over company documents and project data, so that tooling answers from the company's own information rather than from general knowledge.
Work with the Director on the company data warehouse, including ingestion, modeling and the interfaces that tooling queries against.
Maintain source control, environment configuration, testing and deployment practice for everything built.
Build and maintain connectors to the company's platforms using open agent protocols (Model Context Protocol or equivalent), reusable agent skills and scheduled agent runs, with a named service account for every integration.
Treat the company data warehouse as the primary source for every assistant that reports numbers: tooling queries the governed model maintained by the Business Intelligence & Automation Analyst rather than re-deriving figures from raw files.
Maintain a golden test set and regression suite for every deployed assistant. No prompt, model or retrieval change ships without passing it, and a failed check is reported, not suppressed.
Instrument every assistant with tracing and logging so that each run's inputs, retrieved sources, outputs, latency and cost can be reviewed after the fact.
Report monthly, per tool: accuracy against the test set, failure rate, adoption, time saved and API or token spend against the approved budget.
Own production support for deployed assistants: monitoring, failure alerting, incident response, a published support path for end users, and a written runbook for each tool.
Apply and help refine the departmental standards for how tools are built, named, versioned, documented and shared inside the company.
Build the required guardrails into every deployed tool, covering data handling, confidentiality, output accuracy and the boundary between a recommendation and a human decision.
Contribute hands-on testing and written findings to platform and vendor evaluations led by the Director.
Partner with Information Technology on security review, access control, licensing and acceptable use, and raise concerns early.
Classify data before it reaches any model service. De-identify personnel and compensation data, keep client-confidential material inside approved services only, and confirm vendor data-retention and model-training terms before a new service is used.
Route every automated read of company data through named, department-owned service accounts whose activity is logged and reviewed quarterly by the Director and the Executive Vice President.
Verify that generated output meets company brand and accuracy standards before it reaches a client.
Package tools so that an end user can run them without configuring anything, and distribute them to the teams that need them.
Write plain-language documentation for every tool deployed, covering what it does, what it does not do, and when its output should not be trusted.
Sit down with a department and establish what they actually need before building anything for them.
Measure and report adoption and time saved for each deployed tool.
Partner with the company training function on adoption programs, office hours and role-specific training so that deployed tools are used, not just delivered.
Maintain confidentiality of financial, personnel and client information encountered in the course of the work.
Stay current on developments in applied artificial intelligence and bring relevant capability to the department's attention.
Perform other duties and special projects as assigned.