About me

Software Engineer focused on AI-native software development and scalable distributed systems. I draw on 25+ years of hands-on engineering experience to bring clarity to complex problems and guide technical direction. My approach combines thoughtful architecture, pragmatic execution, and mentorship to help teams build reliable software and evolve how they work.

Open to: Staff, Principal, and Distinguished engineering roles, or engineering management.

Follow me:
Brian Raines - Software Engineer

Technical Expertise

AI-Native Engineering

  • Agentic SDLC orchestration
  • AI skill development
  • Model Context Protocol (MCP) server development
  • Codex, Claude, Cursor, GitHub Copilot, Devin, Jules, Amazon Q, Amazon Bedrock

Architecture & AWS

  • Distributed, serverless, and event-driven systems
  • REST and WebSocket APIs
  • Lambda, API Gateway, Step Functions, SNS/SQS, S3
  • Infrastructure as Code: CDK, SAM, CloudFormation

Data Engineering

  • DynamoDB single-table design
  • PostgreSQL, MySQL, MongoDB, Redis, SQLite
  • Data modeling, migrations, and event-driven ETL

Software Development

  • Java, Python, Node.js, PHP, JavaScript/TypeScript
  • Third-party API integration
  • Legacy system modernization

Technical Leadership

  • Architecture and engineering standards
  • Technical strategy
  • Engineering mentorship
  • Cross-team collaboration and knowledge sharing

Quality & Reliability

  • Test-driven development
  • Unit, integration, browser, and end-to-end testing
  • Synthetic monitoring and canaries
  • CloudWatch alarms and Logs Insights
  • AI-assisted root-cause analysis

Experience

—

Principal Engineer, Property Vista — Dallas, TX

Property Vista develops VIDA, an AI-powered multifamily leasing platform connecting renter engagement, tour scheduling, and lease signing.

Organizational Impact
  • Advanced AI-native engineering adoption by mentoring engineers in integrating AI agents into everyday development.
  • Built an operator-invoked Codex skill that traces CloudWatch alarms through Lambda logs, Logs Insights, DynamoDB, and repository code, producing RCA reports with supporting queries and recommended fixes for defects, misconfigurations, and suspected attacks.
  • Built an operator-invoked Codex skill that locates Vivian users by email, phone, name, or web-chat ID and reconstructs system-event timelines and complete conversations from read-only DynamoDB and CloudWatch queries.
Robot Bakery

Developed Robot Bakery using Codex to orchestrate a standard software development lifecycle; continue to maintain and extend it. The Codex-orchestrated agentic SDLC platform is used by all engineers to process Jira tickets through discovery, design, planning, implementation, testing, review, and closure.

  • Paired task-tuned agents with read-only critics; the orchestrator enforces scripted workflow gates and iterates until both agents agree on completion.
  • Enforced test-driven implementation with unit, integration, and browser tests.
  • Enforced human approval of plans before implementation, with operators retaining ownership of pull-request approval, merging, and deployment.
  • Built Pantry MCP, a SQLite-backed persistent memory service with scoped, provenance-aware recall across sessions, tickets, and repositories so agents reuse decisions and lessons.
  • Created repository-based knowledge files documenting missteps and preferred patterns for reuse across developers.
  • Added feature-specific CloudWatch alarms through CDK stacks to monitor deployed application behavior, including API HTTP status codes.
Zero Touch Lease

Contribute to an actively developed proof of concept for end-to-end self-service leasing, connecting Vivian tours, Plaid identity verification, quotes, and lease signing across web chat, SMS, voice, and email.

  • Integrated DocuSeal e-signatures and Payroc application-fee collection; added workflow states for contact and household information.

—

Distinguished Software Engineer, Turnitin — Dallas, TX

Principal Software Engineer | 2021 — 2022

Built cloud-native, serverless platforms at global scale and established architecture patterns and engineering standards adopted company-wide.

Organizational Impact
  • Established AWS SAM serverless patterns as an organizational standard, creating a shared foundation for development.
  • Built cross-team expertise through mentorship in distributed systems, DynamoDB modeling, and event-driven design.
  • Shaped platform technical direction with engineering leadership through strategy and architecture decisions.

Award: 2024 Values Champion Regional Winner: Americas “Action and Ownership, One Team”

Paper to Digital Platform (2022 — 2025)

Developed an AI-powered extension to Turnitin Feedback Studio for paper-based assessments, including handwritten and mathematical responses, using serverless architecture, real-time collaboration, and AI/OCR workflows.

  • Architected serverless, event-driven backends in AWS achieving 99.9% reliability and horizontal scalability.
  • Implemented DynamoDB single-table design and Step Functions for complex AI/OCR workflow orchestration.
  • Built real-time collaboration via WebSocket APIs enabling synchronous grading and feedback.
  • Developed test automation that included unit tests, integration tests, end-to-end tests, and synthetic canaries.
  • Ensured TX-RAMP compliance and certification, maintaining rigorous state-level security standards.

Award: Tech & Learning “Best Tools For Back to School 2024” for Turnitin’s Paper to Digital Add-On for Feedback Studio

Usage Analytics Platform (2021 — 2022)

Designed an event-driven ETL pipeline using SNS, DynamoDB Streams, and Kinesis Data Firehose to feed customer engagement data into Redshift and QuickSight dashboards, supporting consumption-based billing insights across the product portfolio.

—

Software Engineer IV, ExamSoft (Acquired by Turnitin 2021) — Dallas, TX
  • Re-engineered high-volume assessment password management from WordPress to an auto-scaling AWS serverless architecture supporting hundreds of thousands of simultaneous credential requests.
  • Led Liftupp’s migration from a PHP monolith to AWS serverless Java microservices, reverse-engineering APIs and guiding database migration to RDS and DocumentDB.
  • Mentored PHP engineers transitioning to AWS and Java development.

Award: 2020 Engineering Will-to-Win Employee of the Year

—

Lead Software Developer (Consultant), Innovar Solutions — McKinney, TX
  • Architected and delivered a Symfony REST API and Vue.js SPA that unified commercial purchasing and scheduling workflows with real-time data synchronization.
  • Integrated and decommissioned legacy systems, reducing operational complexity and long-term technical debt.

—

Application Architect, Speed Commerce — Dallas, TX
  • Architected enterprise fulfillment and warehouse systems supporting multi-million-square-foot distribution centers and thousands of concurrent users.
  • Designed event-driven order fulfillment and warehouse control systems with real-time integrations to PLCs, sortation equipment, and FedEx, UPS, and USPS.
  • Developed SaaS platforms for global eCommerce operations serving Fortune 500 retailers.

—

Early Career
  • Senior PHP Developer — StoneEagle, Credit Solutions
  • Senior Developer / Engineering Lead — New Media Gateway, Crosswerk, Oven Digital
  • Intranet Coordinator — Garden.com
  • Freelance Web Developer - Siegel Gale, Credit Suisse First Boston, BLUEprint

Built enterprise applications and marketing automation systems and led development teams for Fortune 500 clients including Sprint, Harrah’s, ING, Tiffany, and Consumer Reports.

Education

B.S., Business Administration — University of Texas at Dallas

Projects

Robot Bakery

A Codex-orchestrated development platform that turns Jira tickets into a repeatable engineering workflow. All engineers use it; I continue to maintain and extend it.

Technical notes

My contribution

I developed Robot Bakery using Codex to orchestrate a standard software development lifecycle. Engineers start work with bakery process [jira ticket id]; the orchestrator drives the ticket through discovery, design, planning, implementation, testing, review, and closure.

A workflow with explicit gates

  1. Discovery
  2. Design
  3. PlanningHuman approval before implementation
  4. ImplementationTest-driven development
  5. TestingUnit, integration, and browser tests
  6. Review
  7. Closure

Each step pairs a task-tuned delegated agent with a read-only critic. The primary orchestrator uses codified scripts to enforce workflow state and sends work through further iterations until both agents agree on completion. Agent agreement does not replace human approval: operators approve plans before implementation and remain responsible for pull-request approval, merging, and deployment.

Pantry MCP: context that survives a session

I built Pantry, a local Model Context Protocol (MCP) service and CLI backed by SQLite. It preserves facts, decisions, procedures, and lessons across sessions, tickets, and repositories. Scoped, provenance-aware recall lets agents retrieve relevant context without loading the full history; its search uses local text similarity.

Pantry holds compact operational lessons. Repository-based knowledge files hold detailed, reviewable material about missteps and preferred patterns, shared across developers. Memory entries require accurate scope and provenance and exclude secrets.

Zero Touch Lease

An actively developed proof of concept for self-service apartment leasing, connecting Vivian-guided tours to identity verification, quotes, application fees, and lease signing.

Technical notes

My contribution

I integrated DocuSeal for e-signatures and Payroc for application-fee collection, and added workflow states for collecting contact and household information. These contributions connect payment, signing, and applicant information within the broader leasing journey.

The broader workflow

The proof of concept extends Vivian's self-guided tours into a leasing journey that includes Plaid identity verification, quotes, and lease signing. Vivian supports the workflow across web chat, SMS, voice, and email, with the aim of allowing a renter to complete the journey without human assistance.

Current stage

The thin-thread approach has reached working end-to-end functionality with the integrations connected. Development is active; product hardening is not the current stage. The contribution described here is the integration and workflow-state work I own within that proof of concept.

Paper to Digital

An AI-powered extension to Turnitin Feedback Studio for paper-based assessments, including handwritten and mathematical responses. Developed at Turnitin from 2022 to 2025.

Technical notes

My contribution

I architected serverless, event-driven backends in AWS and built real-time collaboration for the Paper to Digital platform. My work connected distributed systems, AI/OCR workflow orchestration, and synchronous grading and feedback.

Workflow orchestration and collaboration

I implemented DynamoDB single-table design and AWS Step Functions for complex AI/OCR workflows. WebSocket APIs enabled real-time collaboration, supporting synchronous grading and feedback.

Reliability and security

The backends achieved 99.9% reliability and horizontal scalability. I developed test automation covering unit tests, integration tests, end-to-end tests, and synthetic canaries, and ensured TX-RAMP compliance and certification.

Product recognition

Turnitin's Paper to Digital Add-On for Feedback Studio received Tech & Learning's “Best Tools For Back to School 2024” award.

AI Support Skills

Operator-invoked Codex skills for investigating AWS alarms and reconstructing a Vivian user's experience from system records and conversations.

Technical notes

My contribution

I developed two investigation skills that connect operational evidence to an engineer's next decision. A human operator starts each skill in a Codex prompt; the skills use read-only queries and produce reports.

Alarm investigation and root-cause analysis

The operator supplies the CloudWatch alarm contents. The skill queries Lambda logs to locate the error, follows the evidence through CloudWatch Logs Insights and DynamoDB, and examines the repository code that generated the log entry to trace the failure.

The RCA report includes the queries and supporting facts, the most likely cause, and a recommended fix. Investigations can reveal code defects, property misconfigurations, or suspected security attacks. The output is evidence and a recommendation for the engineer, not an automatic production change.

A user's full audit trail

The operator supplies an identifier such as an email, phone number, name, or web-chat ID. The skill first finds the user profile, then follows persisted profile states, transactions, and conversation turns through DynamoDB records and CloudWatch logs.

The resulting report contains a detailed timeline of system events and the complete conversation between Vivian and the user, bringing the relevant records together for review.

What People Say

Contact

Let’s connect. I’m responsive by email and happy to schedule time.

Location

Plano, TX

Resume

Download PDF

vCard

Save Contact