About the Role
We're looking for a Lead Applied AI Engineer/Scientist that will lead the technical design, development, evaluation, and continuous improvement of applied AI systems that solve high-value construction and business problems.
This role is accountable for whether AI agents and machine-learning models achieve defined business targets, not simply whether the technology runs. The Lead Applied AI Engineer/Scientist designs the complete workflow, evaluates performance at each stage, identifies where errors originate, and continuously improves the system until the final output meets established business requirements.
This is a hands-on technical leadership role that sets direction through architecture, experimentation, evaluation standards, and cross-functional influence. The role works across document AI, computer vision, large language models, retrieval, tool-enabled agents, structured outputs, deterministic software, and human-review workflows.
The Lead Applied AI Engineer/Scientist partners with estimators, domain experts, business leaders, data professionals, and software engineers to translate real-world workflows into reliable AI products. Success in this role means AI solutions are measurable, explainable, production-ready, reviewed at the appropriate points, and continuously improved using actual business outcomes.
Core Responsibilities
- Business Problem Definition, Data, and Input Readiness
- Partner with business and domain stakeholders to define use cases, expected outcomes, acceptance criteria, and measurable performance targets.
- Audit real-world inputs, including spreadsheets, vector and scanned PDFs, two-dimensional drawings, scales, symbol conventions, revisions, specifications, images, and three-dimensional models.
- Define the information required at each workflow stage, identify authoritative data sources, and establish how missing, conflicting, or ambiguous information should be handled.
- Identify input-quality issues, workflow dependencies, and technical constraints that may affect model performance or business reliability.
- Design data-labeling and validation approaches with domain experts to establish reliable ground truth.
- Maintain traceability between source information, intermediate outputs, model decisions, and final business results.
- AI Architecture and Workflow Design
- Architect end-to-end AI agents and machine-learning workflows that connect document understanding, computer vision, large language models, retrieval, tool use, business rules, and deterministic code.
- Select the appropriate technical method for each workflow stage based on accuracy, explainability, maintainability, speed, cost, and business risk.
- Define structured outputs, data contracts, tool interfaces, orchestration logic, and failure-handling requirements.
- Design human-review and approval points that allow domain experts to verify outputs efficiently and focus attention on the highest-risk decisions.
- Evaluate build-versus-buy options for models, tools, platforms, and supporting infrastructure.
- Conduct feasibility experiments before major technical or financial commitments and communicate findings clearly to business and technical leaders.
- Model Development, Evaluation, and Error Analysis
- Develop and refine models, prompts, retrieval methods, tools, orchestration logic, and supporting data pipelines.
- Build benchmark datasets and evaluation pipelines that measure individual workflow stages and end-to-end business performance.
- Define meaningful metrics, quality thresholds, test cases, and release criteria based on business requirements.
- Perform detailed error analysis to identify where failures originate and distinguish input, model, prompt, orchestration, tool, and process issues.
- Prioritize improvements based on business impact, frequency, severity, and implementation effort.
- Maintain reproducible experiments, evaluation records, model versions, and technical documentation to support reliable decision-making.
- Pilot Delivery, Productionization, and Monitoring
- Run side-by-side pilots using live projects and compare AI-generated outputs with human work and verified business outcomes.
- Analyze pilot results and explain accuracy, limitations, risks, and improvement priorities in clear business terms.
- Partner with software and data engineers to convert prototypes into maintainable production code, APIs, workflows, and cloud-based infrastructure.
- Establish production-readiness requirements, monitoring controls, fallback procedures, and appropriate human-review mechanisms.
- Monitor production accuracy, failure patterns, user corrections, and changing data conditions.
- Use actual project results and user feedback to improve future model and workflow performance.
- Technical Leadership, Standards, and Responsible AI
- Set technical direction and standards for experimentation, evaluation, documentation, model release, monitoring, and continuous improvement.
- Provide technical guidance, design review, and constructive feedback to engineers, data professionals, vendors, and other contributors supporting AI initiatives.
- Make informed recommendations regarding architecture, models, platforms, vendors, and technical investments.
- Establish appropriate expectations for data handling, access, traceability, human oversight, and responsible AI use.
- Maintain awareness of emerging AI capabilities and assess their relevance to RNGD's workflows and business priorities.
- Balance innovation with production reliability, maintainability, cost, and organizational readiness.
- Cross-Functional Partnership and Business Communication
- Work closely with estimators and other domain experts to understand real workflows, exceptions, review practices, and quality expectations.
- Translate business needs into clear technical requirements and technical performance into practical business implications.
- Communicate model results, tradeoffs, risks, dependencies, and limitations clearly to technical and non-technical stakeholders.
- State candidly when a solution is not yet reliable enough for production use and explain what must improve.
- Gather user feedback, identify adoption barriers, and incorporate practical workflow needs into product development.
- Align technical priorities with business impact while maintaining realistic expectations regarding AI capabilities and delivery timelines.
Preferred Qualifications
- Bachelor's or advanced degree in computer science, machine learning, data science, engineering, or a related field preferred; equivalent relevant experience may be considered.
- Three or more years of experience building applied AI or machine-learning systems delivered to real users, including experience providing technical direction for an AI product.
- Hands-on experience with document AI, computer vision, or complex-document processing, including layout analysis, optical character recognition, object or symbol detection, or measurement from images.
- Experience building large-language-model agents or multi-step AI workflows involving tool use, orchestration, structured outputs, retrieval, and deterministic components.
- Demonstrated experience developing benchmark datasets, evaluation pipelines, error-analysis processes, data-labeling programs, and ground-truth standards.
- Strong Python capabilities and experience collaborating on production software, APIs, data pipelines, and cloud-based AI or machine-learning infrastructure.
- Experience in construction, engineering, architecture, or manufacturing preferred, particularly involving drawings, estimating workflows, vector PDF parsing, geometric reasoning, vision-language models, or historical cost data.
- Ability to explain technical performance, limitations, and recommendations clearly to non-technical stakeholders.
Core Competencies
RNGD Values Alignment
- Exemplifies RNGD's core values:
• Renegade Mentality
• Sweep Floors
• Build Together
• Honor Promises
- Applied AI Architecture and Technical Execution
- Designs reliable end-to-end AI systems by selecting and integrating the models, tools, data, deterministic components, and human-review points appropriate to the business problem.
- Demonstrates a Renegade Mentality by challenging assumptions, testing emerging approaches, and finding practical ways to solve complex problems.
- Evaluation Rigor and Analytical Judgment
- Uses benchmarks, stage-level metrics, production results, and structured error analysis to determine whether systems meet defined technical and business expectations.
- Demonstrates the Sweep Floors mindset by examining the details behind performance, maintaining disciplined documentation, and following failures to their actual source.
- Business Impact and Product Ownership
- Connects technical decisions to measurable business outcomes and prioritizes improvements according to their effect on users, accuracy, risk, cost, and operational value.
- Honors Promises by communicating capabilities honestly, setting realistic expectations, and ensuring released systems perform against agreed standards.
- Cross-Functional Partnership and Communication
- Translates effectively between domain experts, business leaders, data professionals, and software engineers while creating shared understanding of requirements and results.
- Builds Together by incorporating diverse expertise, communicating clearly, and designing solutions with the people responsible for using and supporting them.
- Responsible Innovation and Continuous Improvement
- Balances experimentation and speed with traceability, human oversight, production reliability, maintainability, and responsible use.
- Combines a Renegade Mentality with Honor Promises by pursuing better methods without compromising transparency, quality, or trust.
