Greg B.-Senior Full-Stack Engineer | AI/ML Engineer | AI Evaluator & Trainer

• Integrated LLM-powered capabilities into application workflows for content processing, information extraction, and AI-assisted operations.
• Designed and refined prompts, system instructions, and structured-output requirements to align model responses with product and workflow expectations.
• Built retrieval-augmented generation (RAG) flows using document preparation, chunking, embeddings, and semantic retrieval to supply relevant context to language models.
• Evaluated model responses for instruction adherence, factual accuracy, relevance, completeness, consistency, and formatting against defined criteria.
• Reviewed AI-generated code and technical outputs for logical correctness, edge cases, maintainability, and alignment with stated requirements, where applicable.
• Applied rubric-based assessment and quality-review practices to compare outputs, document defects, and distinguish critical errors from minor response issues.
• Categorized recurring model failures, including unsupported claims, missing requirements, inconsistent formatting, and weak context use, to guide prompt and workflow improvements.
• Contributed to AI training and evaluation workflows through example review, annotation, response grading, and training-data quality checks where included in project scope.
• Validated structured model outputs and integrated AI responses with REST APIs and backend data-processing services, handling malformed outputs and failure paths.
• Architected full-stack marketplace features using React, TypeScript, Java, Spring Boot, and .NET, connecting AI capabilities with production application workflows.
• Optimized PostgreSQL data access and supported AWS-hosted application deployment, troubleshooting, and production reliability.
• Collaborated with product and engineering stakeholders to translate evaluation findings and business requirements into maintainable software and AI workflow improvements.
• Contributed to frontier-model benchmark evaluation by assessing responses across complex instruction-following, multi-step reasoning, and code-generation tasks against explicit task requirements.
• Compared candidate model outputs using consistent scoring rubrics covering correctness, completeness, reasoning quality, constraint adherence, and robustness to edge cases; documented failure patterns to support repeatable evaluation.
• Reviewed benchmark task quality and expected outcomes, including ambiguous prompts, hidden assumptions, and verifier coverage, helping make evaluations more discriminative and reduce false passes or false failures.

TECH: React, TypeScript, Next.js, Java, Spring Boot, .NET, PostgreSQL, AWS, LLM APIs, RAG, embeddings, REST APIs

• Delivered features for a healthcare platform using React, modern JavaScript, Redux, and MobX.
• Implemented Node.js backend functionality and PostgreSQL persistence, coordinating work across UI, API, and data layers.
• Built reusable interface components and validated end-to-end behavior against application requirements.
• Connected application state and API responses to user-facing workflows, keeping interactions consistent across frontend and backend layers.
• Worked with the existing application structure to troubleshoot defects and deliver changes aligned with healthcare product requirements.
• Connected application state and API responses to user-facing workflows, maintaining consistency across frontend and backend layers.
• Troubleshoot application defects and delivered changes aligned with healthcare product requirements.

• Built backend APIs and CMS features for a live-event platform supporting mobile ordering, point-of-sale, reporting, and inventory management.
• Implemented PHP/MySQL business logic and integrations connecting ordering, administration, and operational workflows.
• Developed browser-based administrative features with JavaScript and jQuery.
• Maintained and enhanced existing backend functionality as operational requirements evolved, addressing defects and improving workflow consistency.
• Supported reporting and inventory-related data flows by connecting application logic with MySQL-backed records and administrative interfaces.
• Maintained existing backend features as operational requirements evolved, addressing defects and improving workflow consistency.
• Supported reporting and inventory-related data flows through MySQL-backed records and administrative interfaces.

• Developed client websites and Facebook applications, implementing interactive frontend features and PHP/MySQL-backed functionality.
• Customized site functionality and interface behavior to match individual client specifications.
• Worked across presentation and server-side layers to troubleshoot issues and deliver cohesive web features.
• Customized site behavior to match client specifications and user interaction requirements.
• Worked across presentation and server-side layers to troubleshoot issues and deliver cohesive web features.